Current State of Research on the Upper Layers of the Atmosphere Using Radio Waves1

G. Rukop, Cologne

High-frequency technology makes it possible to study the properties of the upper layers of the atmosphere by various methods. The following possibilities exist here: measurements of field strength, investigations of radio reception, statistics of transoceanic radiotelegraphy, television experiments, special methods of measurement, and direction finding.

Below, data from all these fields are presented; the aim of the present survey is to create as complete a picture as possible of our current knowledge on this question.

I. Earlier Hypotheses and Investigations

Already at the very beginning of the development of radio transmission, unexpectedly large field strengths and, correspondingly, large ranges of wireless telegraphy were observed; these allowed Kennelly (1902) and Heaviside (1902) to put forward the supposition that, at great heights (about 100 km), there existed a conducting layer produced by solar radiation. This same hypothesis had already been expressed by Stewart (1882) and Schuster (1886) on the basis of investigations of the phenomena of terrestrial magnetism. But only considerably later was it possible to provide a large number of experimental confirmations of this hypothesis.

At the present time, the reflecting formation in the upper layers of the atmosphere, which has a more complex character than was formerly supposed, is called the “Kennelly–Heaviside layer.”

In the era of long-wave telegraphy it was not possible to form a sufficiently clear idea of the Kennelly–Heaviside layer. Fluctuations in reception strength and anomalous field strengths observed at large distances were attributed to it. Detailed bibliographic data on this question are given in Saklovsky’s survey. In Section 12 we shall return to this question again.

The discovery of short waves and their application to transoceanic communication provided a great deal of experimental material. The most important results obtained in these investigations are the following:

considerable range of action, often encompassing the entire circumference of the Earth several times over, dead zones (zones of silence), optimum wavelengths varying over the course of the day, with the optimum wave being shortest by day (from 14 to 18 m), somewhat longer at twilight (from 18 to 25 m), and longest at night (from 25 to 40 m).

Further attempts were made at an exact calculation of ionization, ray trajectories, reflection, refraction, and absorption in the upper layers of the atmosphere. They can give satisfactory results only after a sufficient basis has been created for the theory through the experimental study of the remarkable processes occurring in the upper atmospheric layers. We shall speak further on of these experimental results, which at present cannot yet be considered exhaustive.

2. Ray Trajectories

In order to establish the principal points of view on the basis of which the experimental study of the upper atmospheric layers is carried out, it is necessary to make certain assumptions about the basic laws governing the propagation of a ray in an ionized medium (air). Even the simplest assumptions about the creation of the Kennelly–Heaviside layer under the action of solar radiation already make it possible to draw a number of important conclusions. Undoubtedly, at some sufficiently great height, where the density of the air is very small, ionization must be absent. In addition, it is known from measurements that at the Earth’s surface and in the adjacent atmospheric layers ionization is likewise very slight. Therefore the charge density must have a maximum at some altitude, and its course in general must correspond to the curve \(H\), shown in Fig. 1. (Lassen², Becker and Rice³, Pedersen⁴, Försterling and Lassen⁵; in the latter articles extensive bibliographies are given.) Even without precise information on the dependence of the distribution of charge density on height, proceeding only from considerations of the existence of a maximum and of the asymptotic decrease of ionization, one can draw a number of important conclusions about the propagation of rays from a certain source situated on the Earth’s surface.

Fig. 1. The simplest scheme of the Kennelly–Heaviside layer.

Experimental data show that in reality one has to deal with two maxima of the effective charge density (the density distribution is determined by a curve resembling the curves \(H\) and \(T\) in Fig. 9). One may also say that there exist two layers situated at different heights. Owing to this, it is obtained...

there occurs a mutual superposition of the results of reflection of rays from both layers, which, however, can easily be separated into the corresponding parts.

The path of the rays in the case where a single layer exists is shown in Fig. 2, proposed by Fersterling and Lassen. Assuming that a suitable wavelength has been chosen, we obtain the following picture: the rays going most nearly vertically (on the left in Fig. 2) penetrate through the layer; a certain bundle of rays (9 in Fig. 2) travels a long path in the Kennelly–Heaviside layer and is reflected back; the rays closer to the horizontal (1–8 in Fig. 2) are reflected from this layer more quickly. The latter rays can again be reflected upward from the earth’s surface, then again be reflected from the Kennelly–Heaviside layer, and so on (multiple reflection). At first it was supposed that, in transoceanic short-wave radiotelegraphy, the effective rays were rays of type 9. At the present time one must think that these are multiply reflected rays (Fersterling and Lassen[^5]). In favor of this latter supposition speak both calculations of the field strength and the increased reception intensity of transmitters having a horizontally arranged electric vector (Telefunken transradio) as compared with transmitters having a vertical electric vector (owing to better conditions of reflection from the earth’s surface); another confirmation is the regular splitting of the short signal, described in more detail in Section 4. A bundle of rays of type 9 also determines the limiting angle of reflection for a given wavelength and a certain constant concentration of charges in the layer. The quantity determining the limiting angle is the region of maximum effective charge density, i.e. the region \(M\) in Fig. 1. The effective charge density \(N_m\) represents the true density \(N\) divided by the ratio of the ion mass \(m_x\) to the electron mass \(m_o\), i.e.

Fig. 2. Path of rays in the simplest Kennelly–Heaviside layer.

\[ N_m = N \frac{m_o}{m_x}, \]

for more concentrated layers or longer waves, sharper limiting angles are obtained.

3. Statistics of Transoceanic Communication on Short Waves

For transoceanic communication on short waves the most favorable are antennas that give radiation directed tangentially to the earth’s surface, which is explained by the shorter path of the rays and by smaller losses upon reflection. However, for each concentration of the layer, the wave that may be used is the shortest one still capable of being reflected, which proves more favorable owing to the smaller absorption. Because of this, the limiting waves turn out to be variable: namely, at night the longest, at twilight shorter, and by day the shortest; this result agrees with both theoretical assumptions and practical data. For communication, the waves used are chosen, for greater reliability, approximately half an octave longer than the limiting wave. Very detailed and interesting material is given in the article by Queck and Mögel⁶ (an example is shown in Fig. 3), who investigated communication on the routes Berlin—Rio de Janeiro, Buenos Aires, New York, Mukden, Manila, Sydney, Bangkok, Bandung over the course of twenty months at an unchanged wavelength. In some of these routes three different waves were used (daytime, twilight, and nighttime), while in others only two waves. In the above-mentioned work data are also given on the most favorable intervals of time during which multiple reception of signals is obtained best,

Fig. 3. Reception intensities on the New York—Berlin path in 1928 and 1931 (wavelength 20 m).

orbiting around the Earth (the shaded surfaces in Fig. 3), and, as it turns out, there is a difference for the two oppositely directed sections of the great circle (direct and oblique multiple echo). This material contradicts the above-mentioned hypothesis of multiple reflection producing the echo, since the existence of multiple circumnavigation of signals around the Earth on the daytime wave was observed. However, only an exact investigation of the path of the ray can prove that the signals actually passed through the region where night prevailed, and not through a region with a sufficiently large concentration of charges. Another work by Mögel[^7] points to the presence of an interesting parallelism between interference in short-wave communication and fluctuations of terrestrial magnetism (Debye numbers). Here, by interference are meant not the usual atmospheric interferences, but also other disturbances of reception (weakening or disappearance). Recently Mögel[^8] reported that in recent years all limiting waves have become longer (Fig. 3), and that, in order to eliminate the disappearance of reception in the daytime on some sections, it is necessary to use six to eight waves. Here a parallelism is observed with the decreasing activity of the Sun (see also Plendl[^9]). Measurements published by Prescott[^10], made on transoceanic lines, also indicate that at different distances and in different directions there exist different most favorable wavelengths, depending on the time of day and the season.

Fig. 4. Multiple splitting of a simple signal.

4. Investigations in Television

Television was at first built on the same regularities for the most favorable wavelengths that had been established in radiotelegraphy. However, in the investigation of television a very important phenomenon was discovered (1926) (Rukop[^11]), namely—the splitting of a short signal ($S$ in Fig. 4) into several separate components, spaced from one another by approximately 0.001 sec. ($E$ in Fig. 4[^12]). This splitting testifies to the simultaneous existence of several ray trajectories differing from one another in length by 200–300 km; these trajectories can be interpreted as the result of separate zigzag-shaped reflections of various orders (Fig. 5). Owing to this, blurring of reception is produced, as a result of which transoceanic television at the present time appears practically impossible.

In connection with this, let us also mention other interesting results, such as, for example, the Doppler effect and the continuous lengthening of trajectories.

5. Special methods of measurement

Special devices for investigating the properties of the Kennelly–Heaviside layer for the most part relate to nearly perpendicular reflection, since in that case it is possible to place the transmitter and receiver at one and the same location.

The purpose of these investigations is to clarify whether the Kennelly–Heaviside layer, for some definite wavelength, is reflecting, at what height reflection occurs, with what intensity, and how all these factors change over time. On the basis of the above interpretation of the ray trajectory under perpendicular reflection, the limiting waves have the greatest length.

Fig. 5. Zigzag reflection in transoceanic communication.

Some investigators, trying to determine the limiting wavelength, used a continuously varying wave; these investigations required the possibility of simultaneous variation of the wave in the transmitter and receiver. The methods of measurement are divided into two main groups.

A. Interference method

An undamped series of waves is studied under a slow continuous change of wavelength (Appleton and Barnett[^13]). In the receiver, interference occurs between the directly arriving waves (the ground wave) and the reflected waves. The maxima and minima of interference are determined. Hence, with a known change in wavelength, one can determine the path length of the ray and the apparent height of the Kennelly–Heaviside layer.

B. Signal method

In another group of methods, short signals are used, the duration of which is small in comparison with the difference in the travel times of the ground and reflected waves. Since in this case it is necessary to be able to measure a height of about 90 km (travel time \(6 \cdot 10^{-4}\) sec.), the duration of the signal must not exceed \(10^{-4}\) sec. The receiving device must record the difference

in time in such a way that it could be evaluated quantitatively. Various methods are used for this purpose.

1. Oscillograph method. This method was first proposed (1926) by Breit and Tuve[^14]. A signal having the form \(S\) (Fig. 6) first reaches the receiver along the earth’s surface (\(E_0\)); in the presence of reflection it enters the receiver somewhat later once more (\(E_1\)) or even several more times (\(E_2, E_3, E_4\), etc.). From the intervals of time \(t_1 - t_0\) the apparent height can be determined.

Fig. 6. Oscillogram of a signal accompanied by an echo.

In individual measurements the oscillograph also determines the intensity of reception. With the aid of a slit diaphragm the record of the oscillogram can be made point-like (Gilliland and Kenrick[^15]), if synchronized transmitter and receiver are available; this permits continuous registration of reception.

1. Television method. Rukop[^11] proposed in 1926, for measuring the height of the Kennelly–Heaviside layer, to make use of a television apparatus. If, perpendicular to the direction of transmission (the arrows in Fig. 4), a line \(S\) consisting of very short dots is transmitted, then with single reception only one line is likewise obtained. But with multiple reflections several lines arise (\(E\) in Fig. 4). Later this method was also developed for the study of perpendicular reflection, with the recording being made by a rotating glow-discharge lamp with suitable rotating optics (Rukop and Wolff[^16]). In this method, as always in television, control of the transmitter and receiver is synchronized.

Fig. 7. Signal and echo recorded as a Lissajous figure.

1. Measurements with a Braun tube. Goubau and Zenneck[^17] used Lissajous figures, obtained on the screen of a Braun tube, to study reflection from the Kennelly–Heaviside layer. The deflecting field was synchronized with the transmitter, as a result of which the luminous spot on the screen described a circle or another closed figure. The signals were perceived as teeth on the circle (Fig. 7). Their distance determined the time

of propagation, from which it was also possible to compute the apparent height of the Kennelly–Heaviside layer.

An almost analogous method was used by Appleton and Builder^18, and also by Schafer and Goodall^19. Goubau and Zenneck have recently modified their method by using a sweep-frequency unit^20, which made possible continuous recording of the height of the Kennelly–Heaviside layer.

6. Apparent height of the Kennelly–Heaviside layer

All the methods described above make it possible to determine only the apparent, but not the true, height of the layer. Any one of the indicated methods measures only the height at which reflection occurs. Since, however, the reflection of waves of different length occurs in regions of different concentrations characteristic of the given wavelength (for example, \(k\) in Fig. 8), then, with an unchanged geometrical height of the layer but with decreasing concentration \((a, b, c)\), different heights \((h_a, h_b, h_c)\) are measured. Thus the height of the layer may be regarded as equal to the height corresponding to the maximum concentration. Indeed, any limiting wave undergoes reflection precisely in the region of maximum concentration and therefore can be used for these measurements. However, even in this case we are not able to determine the true height of the layer of maximum concentration \(h_{\max}\). The reason for this is that the signal travels a fairly long path in the conducting layer, where it is propagated no longer with the ordinary velocity of light, but with the group velocity (Drude and Sommerfeld^21). Because of this the propagation time is lengthened, and in measurement an exaggerated height is obtained. This difference, as Försterling and Lassen^5 have shown, may reach (when the signal penetrates deeply into the conducting layer) 50%. In what follows we shall consider only the measured apparent heights, since such a calculation is of no fundamental importance.

Fig. 8. Apparent heights of the layer with decreasing charge concentration.

7. Existence of two layers

The results presented here are not arranged in chronological order, since we aim to give here the clearest physical picture of the properties of the Kennelly–Heaviside layer, confirming each stated proposition by experimental data.

Already in measurements of skip zones, and also in measurements of field strength made by Heising, Schelling, and Southworth^22, it was established that medium and especially short waves are reflected from the upper layers of the atmosphere, and, consequently,

one could conclude that the Kennelly–Heaviside layer exists. Quantitative data on reflection were first obtained by special measurements using the interference method in 1925 and by the signal method in 1926. In this way heights were found ranging from 90 km to several hundred kilometers. We shall not dwell on the results of these measurements, but shall turn to more recent data.

Fig. 9. Structure of both reflecting layers at maximum charge density.

The most important newly established fact is the following: there are two layers, situated at different heights, of approximately 100 and 250 km. The thickness of these layers, as well as the effective concentration of charges, is variable, and in its normal state on a summer day is determined by the curves \(H\) and \(T\) (Fig. 9).

The hypothesis of the existence of two layers was put forward by Elias\(^{23}\) as early as 1925. In his opinion, the lower layer, situated at an altitude of 70 km, is produced by the photoelectric action of sunlight, while the higher layer (about 100 km) is due to corpuscular radiation. Although it has not been possible to confirm this hypothesis quantitatively, it undoubtedly contains a share of truth. For the first time, the existence of two layers was experimentally established in 1927 by Appleton\(^{24}\), who also determined the approximate height of the layers. It would be fair to call one of the layers the Kennelly–Heaviside layer, and the other the Elias–Appleton layer.

Fig. 10. Jump of reflection from one layer to another.

In Fig. 10, borrowed from the paper by Appleton and Green^25, are shown the results of a series of measurements that served to establish the existence of two layers. On the curve one can clearly see reflection, continuing for hours, from a layer situated at an altitude of 100 km in the daytime, and from a layer located at a considerably greater altitude at other hours of the day. Reflections from heights of 120–220 km are completely absent. Soon after Appleton, analogous results were obtained (1927) by Breit, Tuve, and Dahl^26. In his later work Appleton^27 reported numerous details of the measurements.

Fig. 11. Double refraction at a wave of 80 m.

The results given below are considered primarily from the point of view of the existence of the “upper layer” and the “lower layer.” It is quite obvious that, according to the above-mentioned law of the limiting wave, the higher layer can be detected only in the case when it is more concentrated than the layer lying below (see also Section 13).

8. Double Refraction

Since we have obtained a large quantity of experimental data, one may try to resolve a number of further theoretical questions. With a considerable change in the brightness of illumination, i.e., in the morning and evening, the concentration of the upper layer has a very large gradient. If observations are conducted at an unchanged wavelength lying in the critical region, i.e., at such a wave which during the observations becomes a limiting wave, then an apparent change in height, disappearance, or appearance of the layer is observed (Fig. 10); this phenomenon has already been explained in Fig. 8.

However, Fig. 10 is not, as we now know, sufficiently complete. Namely, owing to the Earth’s magnetic field, double refraction is produced. This phenomenon, predicted by Heilbert and also by Breit^28, was first quantitatively calculated by Lassen^29. Lassen concluded that appreciable double refraction must arise in those cases when the acting charge is formed by free electrons; on the contrary, it must be immeasurably small if the layer is formed by charge carriers having atomic or molecular mass. The details were expressed by formulas and numbers. Similar formulas were given somewhat later by Appleton and Nesmith^30.

In 1931 double refraction was experimentally found by Appleton and Builder^31, and also by Wolf^32, ^16. Not quite distinct indications of the existence of double refraction were also found by Gilliland^33. Appleton and Builder’s measurements (Fig. 11) clearly prove the existence of two curves caused by double refraction, appearing between 22 and 24 hours, and also between 4 and 5 hours, the refraction being produced in the upper layer. In these measurements the wavelength was 80 m. In Fig. 12 (Wolf’s measurements) double refraction at sunrise is characterized by the branches \(AB\) and \(CB\) (upper layer at a wavelength of 84 m). The appearance of analogous branches of the curve at great heights (about 500 km) is called the “sunrise effect.” It arises in the upper layer quite regularly, for any wave of the critical region, after a more or less short time following sunrise, as is seen in Figs. 18 and 19 (provided only that the upper layer is not accidentally screened by the lower layer). At sunset an analogous phenomenon is also observed, but in the reverse time sequence (Figs. 11 and 13). The reflection splits into two branches, which appear to be high; the heights of both increase, one branch disappears earlier, the other somewhat later (“sunset effect”). It must be noted that double refraction in the Kennelly–Heaviside layer, properly

generally speaking, produce three branches, for example the branches \(DE\), \(GE\), and \(HE\) in Fig. 12, obtained with double reflection and with high resolving power of the apparatus. Schäfer and Goodall \(^{19}\) also

Fig. 13. Double refraction at sunset. Wavelength 84 m.
March 11, 1931.

indicate splitting into three branches (Fig. 17). The splittings observed by Goubau \(^{17}\) will be discussed in Section 11.

9. Normal properties of the layers

A. The lower layer

1. Limiting waves and dependence on time

As regards the effective density of the charges, their apparent height, and their dependence on the time of day and year, we have both regular, often lawfully repeated observational results and a large number of irregular data. Instead of numerical values of the charge density, in what follows we shall indicate the existence or absence of reflection of known waves, or else the limiting waves. At long wavelengths (greater than 500 m) the lower layer, as shown by the observations of Goubau and Zenneck (Fig. 14), is always reflecting. Here attention must be paid to the possibility of multiple reflection, which can lead to an erroneous estimate of the height of the layer, exceeding the true one by two or three times. The circumstance that Goubau and Zenneck found no reflection during the middle of the day, but detected it only after sunset, should not be regarded as an absence of the reflecting layer or attributed to it

Fig. 14. Measured heights of the layer at a wavelength greater than 500 m.

of reduced concentration in the daytime. In all probability, this absence of reflection was caused by absorption, repeatedly observed in various measurements and of fundamental importance for the general investigation of radio-wave propagation (see Section 12).

Fig. 15. Change of the apparent height with change of wavelength.

In 1932 Appleton and Naismith[^30] determined the limiting waves for the lower layer; the result of one separate series of their measurements is shown in Fig. 15, above. With continuous variation of frequency, at the moment when the limiting wave for the lower layer is reached, the reflection abruptly passes from one layer to the other. At the same time an interesting phenomenon is observed, namely: the apparent height of the upper layer and the moment of the jump prove to be greater than for a somewhat shorter wave. This phenomenon should be ascribed to the reduced group velocity of the signal in the lower layer for waves lying close to the limiting wave; it should be noted, however, that this phenomenon has not yet been confirmed by any other measurements.

Fig. 16. Limiting waves of the lower layer.

The magnitude of the limiting wave of the lower layer in daylight for three different months is given in Fig. 16. These values may be regarded as normal values of the limiting waves of the lower layer, from which, however, very large deviations are possible.

1. Heights. As for the height of the lower layer, all observations give one and the same normal value: from 90 to 130 km. For long waves (Gubo and Cennek) the lowest value was found (90 km); for short waves the height apparently exceeds 110 km. The change in the height of the layer at twilight or at night, and also on approaching the limiting wave, is due to causes as yet unclear; this change sometimes proved to be sharply expressed (Appleton^24, Gubo and Cennek^17), while sometimes it was not detected at all (Wolf^32, Paul^35, Schafer and Goodall^19).

B. Upper layer

1. Limiting waves and dependence on time. The upper layer has not yet been subjected to continuous study with respect to its limiting wavelengths, as was done for the lower layer (Fig. 16); however, the results of a large number of measurements at various constant wavelengths allow one to think that the values of the wavelengths for different moments of time can be approximately determined from data on the presence or absence of reflection, in particular when observing the effects of sunrise and sunset.

Fig. 17. Measured layer heights; wavelength 125 m; June 12, 1931, New Jersey.

From Fig. 17, which presents the results of measurements by Schafer and Goodall^19 at a wavelength of 125 m, it is seen that in June this wave was reflected from the upper layer for 23½ hours. The brief disappearance, for only ½ hour, immediately before sunrise proves that this wave, for the most unfavorable time of day in mid-summer, is practically limiting. Wolf^32, and also Paul^34, made numerous measurements at a wavelength equal to 84 m (Figs. 12, 13, 19, 20, 22). They showed, among other things, that when working with this wave in summer it is necessary to consi-

continues with the disappearance of the reflection for 2–3 hours after midnight. An analogous result was found for the wave of 80 m (Appleton and Builder⁸, Fig. 11). Fig. 20 shows a phenomenon that constitutes an exception (see Section 10, A). Gilliland³³ made measurements at a wavelength of 74 m. These measurements showed that in January the reflection is absent after 23 hours.

From the measurements with the 60-meter wave carried out by Paul, it follows that in summer this wave is reflected for many hours, but the height of the layer comes out unexpectedly. In Fig. 18, pp. 492–93, some of these measurements are given, and, among other things, measurements during the solar eclipse of August 31, 1932. The reflection of the 60-meter wave regularly reveals the effect of sunrise and sunset, accompanied by double refraction. In addition, Paul established here the remarkable fact that the charge density first reaches a maximum only in the evening, around 20 hours. Moreover, from the measurements with the 60-meter wave (Fig. 18) it follows that for several hours after noon the reflection is expressed very indistinctly; it is possible that it exists, but is masked by absorption (see Section 12). The influence of the Sun is manifested very sharply. Around 18 hours a reflection with double refraction and a decrease in the height of the layer are observed. Between 23 and 24 hours the effect of sunset appears in the usual form. This maximum of charge concentration between 20 and 21 hours is very interesting, since in these measurements the Sun disappeared below the horizon at about 19 hours. From this it may be concluded that the concentration in the upper layers increases until the disappearance of the last solar rays. However, such an assumption is in contradiction with the usual effect of sunset. It is possible that here one should use the hypothesis put forward by Vegard³⁵ and used by Størmer³⁶ to explain the influence of the Sun’s rays on aurorae, namely: when solar illumination ceases, the upper layers of the atmosphere, which had expanded during the daytime, contract again. The consequence of this is a concentration of charges, producing the evening maximum.

In the phenomenon under consideration the concentration must change in the ratio of approximately 1 : 2, as Lassen showed in an as yet unpublished work; but this ratio is too large for the above-mentioned hypothesis. It is, however, very probable that the “evening concentration” is caused by a stream of charged corpuscles, representing the maximum of the usual corpuscular radiation and reaching our regions after sunset. This phenomenon, as shown by the photographs given in Fig. 18, must regularly recur from day to day. Various meteorological phenomena had already led to an analogous conclusion earlier.

If the above-mentioned maximum is not due to the light radiation of the Sun, then there must exist another, earlier maximum, occurring a short time after noon and corresponding to the maximum equilibrium between radiation and

recombination. Indications of this maximum exist (see below the discussion of the results obtained in measurements with wavelengths of 60, 45, 40, and 35.5 m). A very interesting course was observed on September 8 (Fig. 18). On that day the concentration was apparently very close to the limiting value. The second branch was not observed at all, while the first corresponded to a height greater than 300 km.

During the solar eclipse a wavelength of 60 m was chosen, since at that time it is almost critical for the upper layer, as follows from the photographs shown in Fig. 18. Apparently it would have been less favorable to use a wavelength of 120—150 m, which is critical for the lower layer, as was proposed by Appleton and Chapman^37, since according to our measurements the lower layer in August 1932 had very strong and irregular fluctuations of concentration, as a result of which it would have been difficult to isolate the influence of the solar eclipse, at least if this influence was small. The upper layer, however, owing to its considerably greater regularity, was more favorable for observations. However, as the observations showed, the solar eclipse had only a slight influence. The occurrence of the effect of sunset shortly before 17 h is not unique; it was also observed on other days, for example on September 1 (Fig. 18). On August 31 the effect of sunrise appeared an hour later than could have been expected on the basis of the measurements on the preceding days. However, these slight deviations cannot reliably be attributed to the influence of the solar eclipse. The relatively small number of interferences observed on the day of the eclipse between 17 and 19 h also proves nothing, since the interferences (in this case predominantly station and network interferences, not atmospheric ones) arose extremely irregularly, as was established by measurements on other days.

Only for the evening hours do the above-mentioned results on the reflection of the 60-meter wave disagree with the supposition expressed by Schaefer and Goodall^19. The above-mentioned authors assume that all waves longer than 60 m begin to be reflected soon after sunrise and that reflection is maintained throughout the entire day. However, some of the measurements with a wavelength of 63 m very much resemble the results obtained by Paul with the 60-meter wave (Fig. 18) in that they reveal a maximum of concentration around 18 h.

The same authors, working with a 45-meter wave, found only a short time of reflection (there are no numerical data) at an apparent layer height from 350 to 700 km. Similar results were obtained by de Mars, Gilliland, and Kenrick^38 for a wavelength of 47 m.

At a wavelength of 42 m, despite numerous measurements, neither Wolf^32 nor Paul^34 detected any reflection, which, however, may be explained by insufficient intensity (a 20 W generator). Goubau^39, at a wavelength of 40 m and a power of 6 kW, often observed reflections from the upper layer (in summer, in the daytime). Gilliland, working with a radiated power of 2 kW, found

reflections from the upper layer, having a height greater than 275 km, for a wave of 34.5 m. In these measurements a striking dependence of the height of the layer on the time of day was revealed, contradicting the results of the same author obtained with a 74-meter wave. It is interesting to note that reflection of the 34.5 m wave was restricted only to the months of February and March. A possible connection with compression of the atmosphere or with tides and ebbs in it (Schtermer’s opinion) is unlikely, since the observations were made only during the half-hour before and after noon.

The indications of Kruger and Plendl^40 that the limiting wave at nearly perpendicular reflection is equal to 37 m correspond more or less to the facts set forth above, if one assumes that reflection takes place from the upper layer. On the contrary, the value of the limiting wave, 29 m, indicated by Klepp^41 cannot be compared with the preceding data, since it relates not to perpendicular reflection, but was obtained with a distance between transmitter and receiver of about 90 km. Megel^42 gives interesting results obtained with a 15.6-meter wave (see Sections 10, C and B).

From the above data for the upper layer it follows that the normal changes of concentration as a function of the time of day and the time of year amount to approximately \(1500\%\) (see Section 11).

1. Heights. The data on the normal heights of the upper layer practically coincide for most authors. The lower boundary of reflection for short waves lies at about 220 km. At reduced concentrations or when approaching the limiting wave, the measured heights increase considerably. A height of about 500 km was observed many times, sometimes even up to 700 km. With long waves (400 m), Appleton^27 found a height of 190 km.

10. Anomalous properties of the layers

A. Upper layer

In both layers anomalous concentrations arise; moreover, apparently, anomalous condensations are produced more often, existing for very different intervals of time. In Fig. 19 a short-period oscillation in the upper layer at the moment of sunrise is presented. At this “critical” time the measured height depends extremely sharply on the change in concentration. Similar oscillations are also often observed during the effect of sunset. Continuous recording by means of a glow-discharge lamp makes it possible to detect these irregular changes very well, whereas measurements in which the height is determined by subsequent calculation usually prove insufficiently accurate to allow these oscillations to be distinguished from random errors of measurement. In Fig. 20 a prolonged anomalous condensation of the upper layer is depicted, manifesting itself in the fact that a wave 84 m long was reflected throughout the entire night (May 28), and one

branch existed the entire time, while the other disappeared at about 2 o’clock. A similar course was obtained several days later (June 3–4) by Pauli, working at a wavelength of 60 m. Gilliland, Kenrick, and Norton^43 also obtained, at a wavelength of 60 m, reflection throughout the whole night, the height of the layer reaching 260 km; these measurements were made in September. Since, however, these measurements were not made continuously, but at separate moments in time, considerably distant from one another, I cannot regard them as fully conclusive.

From Fig. 17 (June 12) one may conclude that in mid-summer the limiting wave for the most unfavorable time (3 o’clock) lies at about 130 m. Since the concentrations are proportional to the squares of the limiting frequencies, we have here changes in concentration (in relation to Fig. 20) in the ratio 1 : 4. The course of reflection for a 125-meter wave was also measured in New Jersey, where the “midnight sun” is manifested less sharply than in Cologne, which lies 1000 km farther north. Corpuscular radiation could not have influenced these measurements, since in New Jersey it is in no way less than in Cologne, as may be seen from the map of auroras compiled by Fritz. In New Jersey the normal limiting wave is apparently about 100 m, which was proved by Appleton’s^27 measurements with a 100-meter wave. Thus in this case the change in concentration is still smaller, and the ratio of the concentrations is only 3 : 1. Generally speaking, from all the available measurements one may conclude that anomalous fluctuations of concentration in the upper layer are relatively insignificant and hardly exceed the above-mentioned value of 3 : 1 (see also Section 10, C).

Fig. 19. Oscillations of height after sunrise, wavelength 84 m.

There is evidently no need to speak of anomalous heights of the upper layer. In those cases where anomalous heights were observed, they can apparently be explained by a decrease in the group velocity in the underlying layer (see Fig. 15, above). Exceptional cases are indicated in Section 10, C.

B. The Lower Layer

Anomalous fluctuations of concentration in the lower layer are very large. Significant increases in concentration may occur at any time of day. They are observed in almost every measurement; especially sharp examples are visible in Figs. 15 (below), 18, 21, and 22. The reflected waves visible in Figs. 21 and 22 belong to normal-

for the limiting waves at this moment of time (Fig. 15) approximately as 42:150 and 84:300. Thus the anomalous

Fig. 20. Long-period reflection of waves at 60 m during the course of the entire night of May 28, 1932.

Fig. 21. Anomalous splitting. Strong multiple reflection at night (at the top), weak single reflection corresponds to 120–200 km in height; in the daytime (at the bottom), wavelength 84 m, June 10, 1932.

Fig. 22. Weak multiple reflection at night (at the top), weak single reflection in the daytime (at the bottom), wavelength 84 m, June 10, 1932.

increases in concentration reach at least 1200% (see also Section 10, C).

Anomalous heights of the lower layer were measured by Appleton²⁷ and amounted to 60 km, and on one occasion even 50 km; recently they were also found by Goubau³⁹ (once 50 km), as well as by Gilliland, Kenrick and Norton⁴³ (one observation: 60 km at a wave of 150 m). Toward increased heights, determination of the boundary of the lower layer is very difficult. In the following section, among other things, the corresponding data are given.

B. Indeterminate cases

Observations of almost perpendicular reflection of a 15.6-meter wave are extremely difficult to interpret. Mögel¹⁷˒⁴² observed reflections for several hours at a layer height of about 150 km on various days in winter; once he observed reflection of a 14.9 m wave from a height of 280 km (during a magnetic storm). The height value equal to 150 km is improbable because of the insufficient resolving power of the apparatus used. In all observed cases the reflection was very puzzling, since it has to be ascribed to a layer having a concentration 6–8 times greater than the known concentration; therefore one may think that here partial reflection was observed (see Section 13), in which a large part of the radiation penetrates through the reflecting layer. The same may be said about observations of perpendicular reflection of a 30-meter wave at about 20 hours in January (Mögel⁴²). Sometimes heights lying between 150 and 200 km were observed (Goubau and Zenneck¹⁷, Schäfer and Goodall¹⁹, Wolf, Fig. 21). They were often connected with magnetic storms. In this case, apparently, it is more correct to speak of irregular reflecting clouds than of definite layers.

11. Charge density

From the experimentally found limiting waves one can directly determine, for each case, the maximum charge density of the layer, which for the case of perpendicular reflection is equal to

\[ N=\frac{\pi m_x n_g^2}{e_0^2}. \]

Here \(e_0\) denotes the elementary charge, equal to \(4.7\cdot 10^{-10}\), \(n_g\) represents the frequency of the limiting wave, \(N\) is the number of elementary charges of one and the same type in a cubic centimeter, \(m_x\) is the mass of the charge carrier \((m_0=8.9\cdot 10^{-28})\).

If for a summer day the limiting waves are taken as 35 m for the upper layer and 70 m for the lower layer, then the following possible concentrations are obtained:

\[ N = 9 \cdot 10^{5} \]
(electrons) for the upper layer

\[ \begin{aligned} N &= 2.3 \cdot 10^{5} &&\text{(electrons)}\\ N &= 7 \cdot 10^{9} &&\text{(O-ions)}\\ N &= 14 \cdot 10^{9} &&\text{(O}_{2}\text{-ions)}\\ N &= 6 \cdot 10^{9} &&\text{(N-ions)}\\ N &= 12 \cdot 10^{9} &&\text{(N}_{2}\text{-ions),} \end{aligned} \]
for the lower layer (each possibility is admissible separately, but not together with the others).

Many assumptions have been advanced concerning the causes of the occurrence of both layers; however, all of them have proved insufficient for clarifying this question. The upper layer has a certain, not yet fully clarified, dependence on solar illumination. It must owe its origin to photoelectric action, and the active agents, apparently, are electrons, as follows from the regularly occurring, strongly expressed phenomenon of double refraction (Lassen \(^{29}\)). However, we still cannot say from what gas they are obtained. In the opinion of Försterling and Lassen \(^{5}\), the gas in question is hydrogen or, at any rate, a very light gas, since the very great thicknesses of the layer contradict the assumption of the presence of heavy gases such as, for example, oxygen or nitrogen. It should be noted, however, that in the spectrum of the aurora it has never yet been possible to detect hydrogen or helium lines, whereas oxygen and nitrogen lines have been observed up to heights of 1000 km. Chapman \(^{44}\) assumes that the upper layer is formed by the photoeffect in atoms of monatomic oxygen.

Double refraction in the lower layer is one of the still unresolved problems. Goubau and Zenneck \(^{17}\) repeatedly observed splitting of signals reflected from the lower layer when the distance between receiver and transmitter was only 20–30 km. However, I am not inclined to regard these data as proof of the existence of double refraction, since the wavelength was very far from critical. Moreover, the splitting of the signal occurs irregularly. It is possible that it can be attributed to some cloud-like formation in the layer. Other authors, with the exception of Wolf (Fig. 21), do not give data indicating the presence of double refraction in the lower layer. Thus its existence is doubtful. The absence of systematic double refraction in the lower layer was the reason for the supposition that in the lower layer the active agents are only charges bound to atoms or molecules. However, we still do not have a final solution to this question. The experimental data on double refraction in the lower layer are very contradictory. The lower layer must have a thickness of not less than 30 km (by this thickness is to be understood the distance from the lower edge, having a noticeable concentration, to the maximum concentration, where reflection also occurs). Therefore, to detect

double refraction in the lower layer is considerably more difficult than in the upper layer, which has a thickness of up to 300 km. In any case, the experimental data now available cannot be considered sufficient. However, the assumption of a predominant influence of heavy ions in the lower layer also seems unlikely, since (see the preceding table) in that case there should be only one free electron for \(3 \cdot 10^4\)—\(6 \cdot 10^4\) atomic or molecular carriers. Such a conclusion, however, is very difficult to reconcile with any hypothesis concerning the origin of the lower layer. If it arises under the action of the photoelectric effect, as has repeatedly been supposed, then the free electrons in it must disappear with great speed. But these assumptions lead to recombination coefficients whose order is sharply different from the coefficients known up to the present time and agrees poorly with the duration of the existence of free electrons in the upper layer. In contrast to this it may be supposed that the lower layer is produced directly by ions; moreover, either heavy carriers penetrate directly into it, or else penetrating neutral molecules and atmospheric molecules create atomic charge carriers; the neutral molecules cause normal fluctuations of concentration depending on the position of the Sun, while the charged ones determine anomalous fluctuations of concentration, analogous to the fluctuations during auroras. Undoubtedly, a highly remarkable fact is the coincidence of the height of the lower layer with the region of the most frequent auroras (Størmer \(^{45}\)). Appleton and Naismith emphasize that the normal fluctuations of concentration in the lower layer exceed the fluctuations that might have been expected from changes in the intensity of solar illumination. The action of corpuscular radiation in the lower layer must be very clearly expressed, regardless of whether it will be the only one or whether to it is added the influence of the photoelectric effect, either arising directly in the lower layer or causing the penetration of heavy ions from the upper layers. Unfortunately, the study of auroras gives very little information as to the extent to which the radiation consists of electrons (in auroral rays?) and positive ions (in arcs?) (Størmer \(^{45}\)).

In connection with the questions stated above, it would be very important to determine the dependence of the concentration of the lower layer on geographical latitude. It may be expected that in the equatorial region the lower layer is weakly developed, and that its greatest compression (as well as its irregularities) is formed in the region of the maximum frequency of auroras. Conversely, the upper layer should have a maximum of daytime concentration in the tropics and gradually decrease toward the poles. Measurements of this kind are extremely desirable and should be carried out in the very near future.

Fig. 18. Observations with a 60-meter wave, carried out over

[[unclear: word before “August 4”]] August 4 and September 8, 1932. “Evening concentration.” Solar, August 31, 1932.

Let us mention once again the work of Mögel[^7], who established a complete parallelism between the range of action in short-wave telegraphy and disturbances of the Earth’s magnetism.

To explain this fact one may suppose that magnetic disturbances coincide in time with an intensification of penetrating corpuscular radiation. If the latter produces an anomalous densification of the lower layer, then a radio wave passing across the ocean will undergo reflection already from the lower layer instead of the normal reflection from the upper layer.

Owing to a considerable increase in the total number of reflections over one and the same distance, there will arise an increase in absorption, which will entail a weakening of the field intensity and the cessation of telegraphic communication.

12. Absorption

Under average illumination, i.e. in spring, autumn, and winter, measurements made on various waves almost always revealed multiple reflections from the surface of the Earth and from a certain layer; an analogous phenomenon is also observed on a summer night. In some cases the reflection proved to be sevenfold. Under strong illumination, in particular on a summer day, this phenomenon was absent. Figs. 22 and 23 are extremely indicative in this respect. Fig. 22 represents reflection from the lower layer at different times of the same day. At about 12 o’clock only a single reflection is observed, while at about 2 o’clock it is already sixfold. In the photographs shown in Fig. 23, which refer to the upper layer, at about 15 o’clock (for a wave of 84 m) even simple reflections are absent. At about 18 o’clock they begin to appear occasionally, and at about 20 o’clock multiple reflection arises, continuing until the appearance of the sunset effect. One might suppose that by day the layer proves insufficiently concentrated; but perhaps these phenomena are explained only by an increase of absorption under strong illumination, disappearing with the onset of darkness. This assumption is proved by the fact that the reflection in Fig. 23 is not accompanied by the sunrise effect (cf. Fig. 18), but arises unexpectedly from a normal height. It can likewise scarcely be admitted that absorption occurs when the wave penetrates through the lower layer, since Goubau and Zenneck[^17] discovered the same phenomenon in the reflection of long waves from the lower layer. Therefore absorption should be especially strong in summer throughout the whole day. Appleton[^24], who made an analogous observation, reached this conclusion as early as 1927. An increase in the concentration of each layer with increasing illumination proves insufficient for explaining this question, since one and the same wave will in this case penetrate into the layer to a lesser depth; it is possible that some motion of charges is produced from the lower layer in the direction toward po-

4 Aug.

19 22

8 Aug.

23

\[ \lambda = 60\,\text{m} \]

15 Aug.

31 Aug.

21 23

16

1 Sept.

21

6 Sept.

24

8 Sept.

19 23

between August 4 and September 8, 1932. “Evening concentration.” Solar, August 31, 1932.

surface of the Earth, or else an analogous phenomenon arising independently at lower altitudes (ozone?); however, the quantitative aspect of this hypothesis is still unclear.

Undoubtedly, this absorption is the principal cause of certain remarkable phenomena observed in radio communication; it gives rise to abnormal fluctuations of the field intensity in long-distance communication (on medium-length waves), observed

Fig. 23. Absorption in summer under daytime illumination, wavelength 84 m.

both by day and by night, as well as better transmission on short waves than on medium waves. Conversely, the large range of action of very long waves is due to the stronger ground wave.

Absorption also plays a very important role in short-wave communication. All short waves have the advantage that their absorption is considerably reduced, being proportional to the square of the wavelength. Their range of action is determined only by the reflected rays, while the ground wave has practically no significance. Those waves possess special advantages which, with horizontal radiation, are capable of penetrating through

lower layer, but are reflected from the upper one. Such are the waves used for transoceanic communication (from 10 to 40 m). However, a twofold change in wavelength is already accompanied by a noticeable increase in absorption. Therefore one always tries to work with waves as close as possible to the limiting ones (the change of wavelength depending on illumination).

13. Partial Reflections

Of great theoretical interest is the question of partial reflection. It has already been developed repeatedly, but so far satisfactory results have not been obtained (Seeliger[^46], Gans[^47], Försterling[^48], Epstein[^49], Försterling and Lassen[^45], Gilliland, Kenrick, and Norton[^43]). Measurements suggest that a sharply expressed partial reflection is possible. Simultaneous reflections from both layers, having approximately equal intensities, have been observed. Theoretical calculations allow the possibility of such a phenomenon only for waves lying close to the limiting wave. Of course, the possibility is not excluded that in many practical cases what took place was not optical partial reflection, but that the process was determined by the heterogeneous, cloud-like structure of the lower layer, with breaks, easily penetrable in individual places.

Fig. 24. Multiple echo with a duration of about 0.01 sec.

An interesting hypothesis of partial or diffuse reflection can be found in the work of Eckersley[^50], concerning the echo with a duration of about 0.01 sec., discovered in 1928 (Taylor and Young[^51], Quack and Mögel[^52]); moreover, a five- to sevenfold echo may arise (Fig. 24). By taking bearings on such an echo during reception of signals emitted by the stations Bodmin and Dorchester (Fig. 25 — solid straight lines), Eckersley in Chelmsford (near London) found the directions of the bearings (dashed lines in Fig. 25), which determined the places of formation of this

Fig. 25. Bearing of an echo with a duration of 0.01 sec.

echo at points located at distances of 1000–2000 km and marked by circles in Fig. 25. But these were precisely the places where the radiation from the stations fell into the Kennelly–Heaviside layer, so that it was necessary to assume that at these points scattered radiation arose in the reverse direction. No other explanations of the echo have so far been proposed.

14. World Echo

Hals and Størmer^53 discovered the existence of an echo of very long duration—up to 30 sec. (see also van der Pol, Ellington, Halle^54). Van der Pol^55 attempted to explain it by a decrease in the group velocity; Pedersen^56 objected to this explanation. Størmer put forward the hypothesis that reflection occurs at colossal distances from toroidal electron clouds accompanying the Earth and formed by electrons emitted by the Sun. These same electrons, in Birkeland’s opinion,^57 also account for the occurrence of auroras. A number of data speak in favor of Størmer’s hypothesis. The world echo can be observed on waves that can penetrate both layers in the perpendicular direction (wavelength less than 40 m). In this case the limiting concentration of charges in the toroidal clouds must be greater than in the upper layer.

LITERATURE

1. A. Sacklowski, E.N.T. 3, 438, 1926.
2. H. Lassen, Jahrb. drahtl. Tel. 28, 109, 1926.
3. W. G. Baker and C. W. Rice, Journ. A. I. E. E. 45, 535, 1926.
4. P. L. Pedersen, Dan. Nat. Samf. M. 15, a. b., 1927.
5. K. Försterling and H. Lassen, Z. techn. Phys. 12, 453, 502, 1930.
6. E. Quäck and H. Mögel, E.N.T. 5, 542, 1928; Telefunken-Ztg. 52, 58, 1929.
7. H. Mögel, Telefunken-Ztg. 56, 14, 1930.
8. H. Mögel, Telefunken-Ztg. 60, 32, 1932.
9. H. Plendl, Jahrb. drahtl. Tel. 38, 89, 1931.
10. M. L. Prescott, Proc. I. R. E. 18, 1797, 1930.
11. H. Rukop, E. N. T. 3, 316, 1926.
12. Fig. 4 borrowed from the article by O. Röhm, Telefunken-Ztg. 53, 1929.
13. E. V. Appleton and M. A. F. Barnett, Proc. Roy. Soc. A. 113, 450, 1926.
14. G. Breit and M. A. Tuve, Phys. Rev. 28, 554, 1926.
15. T. R. Gilliland and G. W. Kenrick, Proc. I. R. E. 20, 540, 1932.
16. H. Rukop and P. Wolf, Ztschr. techn. Phys. 13, 132, 1932.
17. G. Goubau, Physik. Ztschr. 31, 333, 1930; G. Goubau and J. Zenneck, Jahrb. drahtl. Tel. 37, 207, 1931; G. Goubau, Ann. d. Phys. 10, 329, 1931.
18. E. V. Appleton and G. Builder, Proc. Phys. Soc. 1, 44, 76, 1932.
19. J. P. Schafer and W. M. Goodall, Proc. I. R. E. 20, 1131, 1932.
20. G. Goubau and J. Zenneck, not yet published.
21. P. Drude, Lehrb. d. Optik; A. Sommerfeld, Ann. d. Phys. 44, 177, 1914.

1. R. A. Heising, J. C. Schelling and G. C. Southworth, Proc. I. R. E. 14, 68, 1926.
2. G. J. Elias, E. N. T. 2, 351, 1925.
3. E. V. Appleton, Nat 120, 330, 1927.
4. E. V. Appleton and A. L. Green, Proc. Roy. Soc. A, 128, 159, 1930.
5. G. Breit, M. Tuve and O. Dahl, Proc. I. R. E. 16, 1236, 1928.
6. E. V. Appleton, Proc. Phys. Soc. IV, 42, 321, 1930.
7. E. L. Hulburt, Phys. Rev. 29, 703, 1927; G. Breit, Proc. I. R. E. 15, 709, 1927.
8. H. Lassen, E. N. T. 4, 324, 1927.
9. E. V. Appleton and R. Naismith, Proc. Roy. Soc. A 137, 36, 1932.
10. E. V. Appleton and G. Builder, Proc. Pys. Soc. A 44, 76, 1932.
11. P. Wolf, Figs. 12, 13, 19, 20, 21, 22 from the article published in Jahrb. drahtl. Tel.; see also 16.
12. T. R. Gilliland, Proc. I. R. E. 19, 114, 1931.
13. E. Paul, Figs. 18 and 23, from as-yet unpublished measurements made at the Institute of Technical Physics of the University of Cologne.
14. J. Vegard, Z. Physik 16, 367, 1923, see also 36.
15. C. Störmer, Ergebn. d. Kosm. Phys. 1, 1931, A. V. G. Leipzig.
16. E. V. Appleton and S. Chapman, Nat 129, 757, 1932.
17. P. A. de Mars, T. R. Gilliland and G. W. Kenrick, Proc. I. R. E. 19, 106, 1931.
18. G. Goubau, E. N. T. 10, No. 2, 1932.
19. K. Krüger and H. Plendl, Jahrb. drahtl. Tel. 33, 85, 1929.
20. J. K. Clapp, Proc. I. R. E. 17, 79, 1929.
21. H. Mögel, Telefunken-Ztg. 60, 29, 1932.
22. T. R. Gilliland, G. W. Kenrick and K. A. Norton, Proc. I. R. E. 20, 286, 1932.
23. S. Chapman, Proc. Roy. Soc. A, 132, 353, 1931.
24. C. Störmer, cited work, 36, § 16.
25. H. Seeliger, Phys. Ztschr. 5, 237, 1904.
26. R. Gans, Ann. d. Phys. 47, 709, 1915.
27. K. Försterling, E. N. T. 5, 530, 1928.
28. P. S. Epstein, Proc. Nat. Acad. Sc. 16, 627, 1930.
29. T. L. Eckersley, Journ. A. I. E. 67, 992, 1929.
30. A. H. Taylor and L. C. Young, Proc. I. R. E. 16, 561, 1928; 17, 1491, 1929.
31. E. Quäck and H. Mögel, E. N. T. 6, 45, 1929.
32. See C. Störmer, Nat 122, 681, 1928; Naturwiss. 17, 643, 1929.
33. B. v. d. Pol, E. V. Appleton and J. R. Galle, cited work, 36, § 23.
34. B. v. d. Pol, Tijdschr. Ned. Rad. 4, 13, 1928.
35. P. O. Pedersen, Proc. I. R. E. 17, 1750, 1929.
36. E. Birkeland, see 36.

1. Elektrische Nachrichten-Technik (E. N. T.) 100, No. 2, 1933; translated by N. N. Malov. ↩