Meteor Ionization and Ionospheric Anomalies*

A. Lovell

Introduction

When a meteor enters the Earth’s atmosphere, it creates an ionized trail at an altitude of about 100 km above the Earth’s surface. Under suitable conditions these ionized clouds have a sufficiently high coefficient of reflection of radio waves for a brief reflected signal to be received on a sensitive receiver. The development of radar techniques in recent years has created a method for studying meteors by day and by night, a method that possesses greater sensitivity than the human eye and that has already begun to give us new data on the physics of the atmosphere at these altitudes. There are reliable experimental indications that the overall effect of meteor ionization is also one of the principal causes of the more prolonged anomalies in the ionospheric \(E\) layer.

The present review is divided into two main parts: the first part is devoted to the history of the study of anomalies in the ionospheric \(E\) layer and to a discussion of present views on the nature of the sporadic \(E\) layer; in the second part, contemporary studies of brief reflections from meteor trails and the applications of these studies to meteor astronomy and to the physics of the upper layers of the atmosphere are considered.

* Reports on progress in physics (Phys. Soc.), Vol. XI (1946–1947), London, 1948, pp. 415–444. Somewhat revised translation.

In view of the fact that A. Lovell’s article contains many unnecessary references, often to papers by one and the same author that contain identical results and were merely published in different journals, as well as to papers containing obsolete and relatively insignificant data, the editors have shortened the list of cited literature and supplemented it with references to later publications for greater ease of use. —Editors.

I. ANOMALOUS EFFECTS IN THE \(E\) LAYER AND THEIR CONNECTION WITH METEORS

1. General information on the structure of the ionosphere

The principal features of the structure of the ionosphere were established by studying the reflection of radio waves from the upper layers of the atmosphere, chiefly by the pulse method. In this way the existence of two main ionized layers was established—the \(E\) layer at an altitude of \(120\) km and the \(F\) layer at an altitude of about \(250\) km. Much information has been obtained about these layers of the ionosphere, and extensive reviews on this subject have been published\(^{1,2,3,4,5}\).

For the purposes of the present review it is sufficient to note that, under normal conditions, the electron density in the \(E\) layer shows a distinct connection with the Sun, reaching a maximum at noon and a minimum (owing to recombination and attachment) during the night. For example, in England at noon on a summer day the density in the \(E\) layer is about \(1.1\cdot10^{5}\) electrons in \(1\ \mathrm{cm}^{3}\), decreasing by midnight to approximately \(7.9\cdot10^{3}\) electrons/\(\mathrm{cm}^{3}\). The equivalent critical frequencies of radio waves are \(3\) Mc and \(0.8\) Mc, respectively. Waves of higher frequencies pass through the \(E\) layer without being reflected. The mean electron density in the \(F\) layer is about \(6\cdot10^{5}\) electrons/\(\mathrm{cm}^{3}\), corresponding to a critical frequency of \(7.0\) Mc.

2. Discovery of anomalous effects in the \(E\) layer

An indication that ionization in the \(E\) layer does not decrease smoothly during the night was obtained in 1928 by Heisig\(^{6}\). He used pulse apparatus at frequencies of \(2.7\) and \(5.2\) Mc to study the height of the ionized layers. After sunset, sharp decreases in the height of the \(E\) layer were observed, corresponding to sharp increases in the electron density in this layer. These results are illustrated in Fig. 1, which shows that the normal tendency of ionization in the \(E\) layer to decrease during the night is accompanied by separate abrupt intensifications of it.

The author writes: “The impression is created that a large number of electrons are introduced rather rapidly into the atmosphere, as a result of which the layer descends, and this descent is accompanied by turbulence and by fluctuations of density near its lower boundary.” He concluded from this that the ionization conditions after sunset are unstable and do not change in a regular manner; he supposed that the cause of the irregularities might be particles coming from the Sun.

Further indications of anomalies were obtained by Eckersley\(^{7}\). The propagation of radio waves in the frequency range from \(6\) to \(20\) Mc was studied. It was found that, under conditions in which a transition of reflections from the \(E\) layer to the \(F\) layer was observed, the signals were often reflec-

... occurred in the intermediate zone. The author attributed this effect to the presence of dense ionic clouds below 125 km, but did not discuss their nature.

Other authors who studied anomalous nighttime effects came to the conclusion that “either something impedes the recombination of ions, or there exists some ionizing agent that can act on the dark side of the Earth,”⁸ and that the rapid change of ionization during the night from one moment to another indicates the presence of some ionizing agent of non-solar origin.⁹

3. Initial views on meteors as the cause of anomalous effects in the layer \(E\)

The suggestion that meteors may create sufficient disturbances in the layer \(E\) to influence the propagation of radio waves was first expressed by Nagaoka,¹⁰ who believed that a meteor would “sweep out” electrons along its path and, although it would itself ionize the air, the number of electrons it created would be small compared with their number present before the meteor passed. Thus, there would be fewer electrons on the meteor’s path than in the surrounding air, and this would create abrupt changes in the refractive index for the incident radio wave.

Fig. 1. Change in the height of the layer \(E\) (circles indicate measurements in which multiple reflections were observed).

The suggestion that the ionization itself caused by meteors may influence the conditions in the layer \(E\) was made in 1931 by Skellet,¹¹*) and it was indicated that the results obtained earlier⁶ could be explained in this way. His theoretical estimates of the magnitude of the ionization produced by meteors were based on Merris’s theory,¹² and at the present time it is considered¹³ that they are erroneous even by an order of magnitude. Nevertheless, the experimental

) This suggestion was first made by N. A. Ivanov,¹¹ᵃ who presented numerous arguments based on consideration of the conditions for the propagation of radio waves during the night. Editors.*

studies carried out during the Leonid shower of 1931\(^{11,9}\) gave convincing indications that meteor ionization is responsible for some of the nocturnal anomalies in the \(E\) layer.

In these studies a pulse method was used at frequencies between 1.6 and 6.4 MHz, the frequencies being rapidly alternated. During the Leonid shower of 1931, strongly disturbed conditions with sharp anomalous increases in ionization were found in the \(E\) layer, reaching a maximum on the night of November 16–17, which, as is known from visual observations, corresponds to the maximum of the shower. Unfortunately, on these nights there were magnetic disturbances, and therefore it could not be definitively established that the effects were caused by meteor ionization. However, during the Leonid shower of 1932 successful visual correlations were obtained between meteors passing overhead and sharp short-lived increases of ionization in the \(E\) layer\(^{11}\).

Fig. 2. Short-lived increases in ionization of the \(E\) layer associated with the passage of meteors overhead.

In Fig. 2 these sharp increases in ionization are shown together with data on the meteors that passed by. On a night when there were no active meteor showers, these sharp increases in ionization proved to be very rare—an analogous graph for the night of February 19–20, 1929, shows only one such increase.

From these studies the following conclusions were drawn: a) nocturnal increases of ionization in the \(E\) layer are most sharply expressed during meteor showers; b) in all cases of the sharpest increase of ionization, meteors passing overhead were observed; c) intermittent reflections lasting only a few seconds are rare, except during meteor-shower periods; d) observations of the critical frequency showed that during the Leonid maximum the ionization in the \(E\) layer reaches \(10^6\) electrons/cm\(^3\), i.e., more than at summer noon.

In Japan\(^{14}\) the influence of the Leonid shower of 1932 was also investigated, and a considerable increase in the number of short-lived echoes during the shower was found, but, apparently, individual echoes were not linked with meteors.

Even before these investigations, disturbances on long-distance shortwave communication lines had been specially examined with the aim of establishing the existence of a correlation with meteor streams15, 16. A certain correlation was found, but because of the high sensitivity of such communication lines to magnetic disturbances, the analysis did not yield reliable results.

4. Long-lasting anomalies (sporadic layer E)

The works described above showed that the intrusion of a meteor into the Earth’s atmosphere can create a short-lived echo from the vicinity of the E layer. However, these works did not indicate what role meteor ionization plays in long-lasting anomalous increases of ionization, and from the experimental data it was not possible to draw any conclusions in this respect, apart from the general conclusion that nocturnal increases of ionization in the E layer appear most sharply during meteor streams.

For several years after the discovery of long-lasting anomalies, many hypotheses were put forward concerning their cause, and it is most convenient to consider the various proposals separately.

A. Thunderstorms and barometric effects. Daytime observations of anomalous increases of ionization in the E layer, between May 1931 and June 1932, led to the suggestion17 that they are accompanied by peculiar isobaric situations characterized by the presence of low-pressure areas at the observing site or to the north of it. Several years before this, the suggestion had already been made that thunderstorms could cause ionization of the upper layers of the atmosphere. Observations by Appleton and Naismith (January—July 1932) showed18 that the correlation coefficient between the ionization density and the “thunderstorm index,” based on observations of atmospherics, is approximately 0.75.

The thunderstorm hypothesis received further support in subsequent works, from which it followed that nocturnal increases of ionization occur at a constant height of 105 km (see19), and also that daytime sporadic echoes, lasting an hour or two, occur at this very same height. Analysis of the data showed that 74% of thunderstorm days exhibit nocturnal anomalies in the E layer, whereas among days without thunderstorms this effect occurs on only 46% of days.

All these works showed that the agent responsible for anomalous increases of ionization affects only the E layer and does not affect the F layer.

The conclusion later expressed in the literature, that thunderstorms are responsible for maintaining ionization in lower reflecting layers of the atmosphere, proved untenable, since the existence of these lower layers was not subsequently confirmed.

Measurements made in Calcutta showed a sharp correlation of the anomalous layer \(E\) with thunderstorms. In Australia a correlation was found between the presence of nighttime anomalies in the \(E\) layer and the barometric pressure at the Earth’s surface observed on the following morning.

However, subsequently (in 1937–1938) a number of authors established the absence of correlation between the anomalous layer \(E\) and thunderstorms and came to the conclusion that the arguments in favor of a connection between the anomalous layer \(E\) and thunderstorms were insufficient, and also that the barometric effect found in Australia did not exist in England.

From that time the idea that thunderstorms and barometric effects play a noticeable role in anomalous increases of ionization in the \(E\) layer was abandoned. Owing to the imprecision in establishing the thunderstorm index and the area over which storms could be considered effective, such connections were very difficult to study, and it was necessary to accumulate statistical data covering a period of several years in order to be able to establish the erroneousness of the original idea.

B. Magnetic storms and auroras. In 1929 observations were carried out during a magnetic storm, and a noticeable change in the height of the \(E\) layer and in the character of the echo was established[^20]. The possibility of a connection between anomalous ionization in the \(E\) layer and magnetic activity was studied by Appleton and Naismith in 1932[^8]. In 1933 the same authors found a frequent, though not invariable, connection between these phenomena, but an increase in ionization of the \(E\) layer was always observed during magnetic storms.

In the expedition during the International Polar Year (1932) this connection was investigated in detail[^21]. Although a distinct connection was revealed between magnetic activity and anomalous ionization of the \(E\) layer, the character of the connection proved complex. The general idea was that the intrusion of charged solar particles into the upper layers of the atmosphere should create an increase in ionization and, therefore, it was to be expected that anomalous ionization would be observed at the same time as magnetic storms and auroras.

However, in some cases, when the magnetic activity was exceptionally great, the echo from the \(E\) layer disappeared. To explain this fact it was proposed that when solar particles are especially numerous and penetrating, they also ionize the lower layers of the atmosphere and thus create an absorbing screen.

Figure 3 shows a graph which, in a surprisingly clear form, shows the connection between the anomalous increase of ionization and a small magnetic disturbance.

The results of the Polar Year expedition showed a distinct connection between magnetic disturbances and anomalous ionization of the \(E\) layer, but the complexity of the phenomena was such that many later investigators did not agree with these results. Some

which of them obtained data confirming the results of the polar year, whereas others found no correlation. Anomalies were found both on magnetically disturbed and on quiet days, and it was also established that the anomalous layer \(F\) is associated with sharp magnetic disturbances and auroras.

Fig. 3. Change in the anomalous layer \(E\) in Tromsø during a small magnetic disturbance on March 22–23, 1935.

Radio fadings in shortwave communication, as is now known, precede similar anomalies, but nevertheless are connected with them. In 1935 attention was drawn to the frequent recurrence of this phenomenon, but the general connection of disturbed conditions of radio-wave propagation with solar and magnetic disturbances had already been discussed earlier—as early as 1927. These effects are apparently connected with the conditions of echo disappearance found during the polar year, and are caused by solar particles ionizing the layers lying below layer \(E\) and, in this way, creating an absorbing layer that excludes the possibility of normal reflection of radio waves from layer \(E\). A recent study \(^{22}\) showed a connection between solar eruptions, solar radio emission, magnetic storms, and similar radio fadings.

V. Meteors. For a number of years after the work of Skellett \(^{11}\) and Schafer and Goodall \(^{9}\), attempts to connect ionization in layer \(E\) with meteors

were carried out only in India*). During the Leonids of 1933 the electron density in layer \(E\) was measured \(^{24}\) (Fig. 4). On the nights of November 13 and 14 the electron density reached values of \(3.3 \cdot 10^5\) and \(2.2 \cdot 10^5\) electrons/\(\text{cm}^3\), respectively. The investigators note that such high night-time densities are exceptional and had never been recorded by them during prolonged observations in the polar year (1932–1933). According to records from nearby observatories, these days were free from magnetic or solar disturbances.

Fig. 4. Electron density in layer \(E\) during the Leonid shower of 1933.

Similar measurements were carried out during the Leonids of 1936 \(^{25}\). These measurements also indicated a clear connection between the increase in night-time ionization and the maximum of the shower. Searches for effects in layer \(F\) gave a negative result.

During the Draconid meteor shower in 1946, ionograms were obtained indicating that this meteor shower caused the appearance of a reflecting layer which existed for several hours at a frequency of \(3.5\) Mc \(^{26}\).

Recently, convincing evidence has been obtained in favor of the view that part of the anomalous ionization of layer \(E\) is connected with meteors. The number of short-duration echoes at a frequency of \(27\) Mc was compared with the appearance of the sporadic layer \(E\). The results of measurements during the intense but short-lived Draconid meteor shower in October 1946 are shown in Fig. 5, from which it is evident that the correlation between the sporadic layer \(E\) and the frequency of occurrence of meteor echoes is very close. The annual variation of short-duration echoes and of the sporadic layer \(E\) is shown in Fig. 6. The agreement here is also quite satisfactory.

c. Structure of the anomalous layer \(E\). In Section 2 it has already been mentioned that Eckersley \(^{7}\) came to the conclusion that the short-duration scattering observed in the transition of reflections from layer \(E\) to layer \(F\) is due to night clouds below \(125\) km. A similar supposition—that the anomalous ionization of layer \(E\) may be the result of the presence of clouds with increased electron concentration—has also been expressed by other authors—

* In the USSR, M. N. Gnevyshev \(^{23}\) established the influence of the Draconid meteor shower of 1933 on the propagation of radio waves. —Editor.

... by radars^28, which measured reflected signals from the \(E\) layer at distances up to 200 km on waves in the range 200–500 m. The presence was established of scattering centers with an effective scattering radius of about 200 km. There were indications in the literature that the anomalous effects extend over a region with a transverse dimension of about 100 km. Some investigators believed that the anomalies represent a case of partial reflection at a sharp boundary, but subsequently a number of authors examined this question in detail and came to the conclusion that under anomalous conditions the \(E\) layer contains clouds with an electron density exceeding the density in the surrounding regions, and that the anomalous ionization is not layered in character, but is caused by the accumulation of such clouds, immersed in the \(E\) layer, with an ionization maximum 7 km below the maximum of the normal ionization of the \(E\) layer. The existence of such scattering clouds was independently proposed on the basis of a study of short-lived scattering centers.

Fig. 5. Changes in the number of ionization “bursts” \((A)\) and of sporadic ionization of layer \(E\) \((B)\) during the night of the Draconid meteor shower (curve \(C\) shows the number of ionization “bursts” on an ordinary night).

Fig. 6. Annual variations in the number of ionization “bursts” and in the frequency of occurrence of the sporadic \(E\) layer in the daytime (average for 1944–1946).

D. Possible causes of anomalies in the \(E\) layer. Experimental data indicate that there are two principal causes of long-lasting anomalies in the ionization of the \(E\) layer:

a) increases of ionization associated with magnetic activity and caused by solar corpuscles;

b) increases of ionization caused by meteoric ionization. Such ionization, under suitable conditions, may be detected...

detected as a short-lived echo at a frequency close to the critical one. The accumulation of these effects may produce prolonged anomalous ionization.

Owing to the deflection of solar corpuscles in the Earth’s magnetic field, the first cause, as might be expected, should predominate at high latitudes. Therefore one can readily explain the differences in the results obtained by various investigators, in particular the clear correlation with meteor activity found by Indian investigators, and the absence of correlation of the anomalous layer \(E\) with magnetic activity found by many investigators working at low latitudes. It is also possible to explain the changes of anomalous ionization with latitude.

5. Short-lived anomalies

In Section 3 the first investigations of short-lived anomalies in the \(E\) layer were described. The anomalies were associated with ion clouds below 125 km, and it was shown \(^{9, 11}\) that they are caused by meteor ionization. However, this conclusion was by no means generally accepted, and up to 1940 no direct indications had been obtained of a connection between short-lived echoes and meteors. In the present section we consider information on these short-lived scattering centers obtained during this intervening period.

Short-lived echoes were also observed during the polar-year expedition \(^{21}\), the authors referring to meteor ionization as a possible explanation. In 1932 Eckersley \(^{7}\) made an attempt to give a theory of the scattering of radio waves by ion clouds in the \(E\) layer (see Part II). However, in discussing the nature of the scattering centers he did not go beyond indicating that ionization in the \(E\) layer must be irregular in space and in time. In 1937 he obtained a photographic recording of echoes from short-lived clouds, using a pulse transmitter at 9.1 MHz, and found that short-lived reflections of duration of the order of a second are observed almost continuously, both by day and by night. Among other suggestions, the author mentions that these echoes may be caused by “small meteors.” These results constitute strong support for the hypothesis of the meteor origin of the ionization \(^{29}\).

Calculations of the equivalent nocturnal density from registragrams of ionization bursts at distances between 80 and 100 km at a frequency of 6 MHz showed \(^{30}\) (1937) that for several seconds it should be of the order of \(10^{12}\) ions/cm\(^3\).

The first detailed measurements of short-lived echoes were carried out in 1937 \(^{31}\) at a frequency of 8.8 MHz. The obtained distribu-

tion of heights is shown in Fig. 7. These authors believed that the echoes could be divided into two groups, one consisting of echoes with durations from fractions of a second to 1–2 seconds, and the other of echoes with durations from several seconds to several minutes. From the amplitude of the echoes it was possible to conclude that the scattering centers are clouds containing \(10^{16}\) electrons, concentrated in a region whose linear dimensions are small in comparison with the wavelength. It was suggested that these results indicate the penetration into the atmosphere, by day and by night, of some agent producing bursts of ionization of very considerable intensity.

Fig. 7. Distribution of heights of short-duration echoes (measurements of 1937).

In 1940, Eckersley carried out a study of echoes by analyzing the scattering of radio signals with frequencies from 7 to 18 Mc/s, under conditions of passage of the reflection \(^{37}\) (see Section 2). The results of this study with respect to the distribution of heights, diurnal variation, and distribution of durations are given in Figs. 8, 9, and 10. The distribution of heights agrees well with the distribution shown in Fig. 7.

Fig. 8. Distribution of heights of short-duration echoes (observations at wavelength 39.5 m from August to November 1936).

The author does not discuss the nature of these scattering clouds, but comes to the conclusion that a continuous transition is possible between these short-duration echoes and prolonged anomalies in the \(E\) layer.

Later Eckersley and Farmer \(^{32}\) measured the direction and polarization of waves reflected from short-duration clouds with the aid of two antennas placed at some distance from one another.

Large changes of phase over fractions of a second were detected, which indicated changes in the direction of the source by as much as \(15^\circ\) in \(1/4\) second. The authors concluded that the echoes could not be caused by individual clouds, but were produced by numerous centers distributed in the \(E\) layer, or else by a swarm of particles flying into the atmosphere, and consequently that they could not be caused by meteoric ionization. However, earlier work showing that the echoes are caused by meteors has now been fully confirmed, and in light of modern investigations (see Part II) the results of Eckersley and Farmer can be interpreted in terms of the meteoric nature of the echoes.

Fig. 9. Count of the number of echoes on the \(39.5\)-m wave in observations of October 15–16, 1938.

In Norway\(^{33}\), short-duration echoes at a frequency of \(6\text{–}8\) Mc/s were studied. A height distribution was obtained, analogous to those shown in Figs. 7 and 8; the durations of these echoes range from \(0.5\) to \(2\text{–}3\) seconds; no echoes with duration \(1/10\) second or less were detected. In discussing the nature of short-duration echoes the author did not go beyond the assertion that they are produced by clouds of electrons.

Fig. 10. Distribution of echo durations on the \(39.5\)-m wave (from August to November 1936).

During the 1940 Leonid stream, short-duration echoes were observed at a frequency of \(3\) Mc/s, while meteors were flying overhead\(^{34}\).

The measurements of short-lived signals described above were made either directly, using pulse techniques, or by means of observations of scattering under conditions in which the frequency was sufficient for penetration through the normal \(E\) layer and reflection from the \(F\) layer. There exists a third type of phenomenon, which also may be caused by the presence of short-lived dense ion clouds in the \(E\) layer—the propagation of radio signals over a great

Fig. 11. Diurnal variation in the number of “bursts” and meteors.
\(a\)—mean number of bursts per hour; February 1943–January 1944;
\(b\)—number of meteors per hour according to Schmidt’s observations, \(c\)—the same according to Cuvier-Gravier’s observations, \(d\)—the same according to Hoffmeister’s observations.

distance in cases where the frequency is such that both the \(E\) layer and the \(F\) layer are penetrated, and the signal is normally not received beyond the region of propagation of the ground wave. As early as 1928–1929 there were reports of anomalous scattering effects, in which high-frequency radio signals were received at a distant station after traversing a very long path. In 1931 individual signals were detected at very great distances on as high a frequency as 40 Mc/s. In 1933 bursts of signals were observed at distances of up to 200 miles from transmitters operating in the frequency interval from 36 to 100 Mc/s, and individual maxima were of very great intensity. In 1938 phenomena of this type were examined, and it was pointed out[^35] that the bursts could be due to meteoric ionization.

In 1942–1944 these bursts were studied in detail, using transmitters at frequencies from 42 to 84 Mc/s and observations at distances from 100 to 340 miles[^36]. During August and November 1944, in several cases visual correlations were obtained between bursts and meteors. The diurnal variation of bursts (Fig. 11) shows good agreement with theoretical

and experimental data on the diurnal variation in the number of meteors. It was established that at a frequency of 71.75 MHz the bursts occur less often and are of shorter duration than at a frequency of 44 MHz.

6. Echoes from heights exceeding the height of layer E

Many investigators of short-duration echoes sought analogous effects occurring in layer F, but with negative results, and Bkharr^25 came to the conclusion that meteors do not exert any measurable influence on the ionization of layer F. Echoes with effective heights greatly exceeding the height of layer F were explained by the fact that the signal passes through layer E, is then reflected from layer F, reflected upward from layer E, reflected a second time from layer F, and then penetrates through layer E to the Earth—an echo of “type M.”

Echoes from heights of 600 to 1800 km, which were reported in 1934, also apparently are echoes of “type M,” or else may be interpreted as the result of lateral scattering from short-duration clouds in layer E. In 1940 similar effects were observed, and new data were obtained in favor of the view that these echoes are due to scattering by short-duration clouds in layer E, with the scattered signal returning to the Earth after reflection from layer F^37. Apparently, anomalous ionization in layer E can give a satisfactory explanation of most observations of echoes from large apparent heights. Some time ago rapidly moving ionospheric clouds were allegedly discovered, moving from a height of 800–900 km toward layer E and sometimes returning back^38. The authors attributed these clouds to streams of solar corpuscles, but in the data they published there is nothing that would exclude this effect from the category of those considered above.

Recently^39, at a frequency of 46 MHz, an echo was discovered from a luminous auroral cloud at a height of 480 km, which persisted for 30 minutes.

II. MODERN INVESTIGATIONS OF SHORT-DURATION METEOR ECHOES

The development during the war of radio transmitters and receivers operating at frequencies considerably exceeding the critical frequencies of layers E and F gave impetus to the study of short-duration meteor echoes. The investigation by Hay and Stewart^40, in which frequencies of the order of 60 MHz were used, showed that, beyond any

doubt, at least some of these short-duration echoes at high frequencies are associated with meteors entering the earth’s atmosphere.

During the last two years, significant progress will be made in the application of this technique to meteor astronomy and to the physics of the upper layers of the atmosphere.

7. Application to Meteor Astronomy

A. Meteor activity. Until now, the obtaining of information about meteor activity has depended on the darkness and clarity of the sky. Radio engineering removes both of these limitations and makes it possible to conduct a continuous study of the level of meteor activity.

At present it is known that the number of meteor echoes observed by radio depends very strongly on the wavelength.

Fig. 12. Simultaneous observations of the number of short-duration echoes at frequencies of 36 and 72 MHz.

At wavelengths less than 6 m the number of echoes is very small, except during periods of meteor showers, and there is a close correlation between these echoes and visually observed meteors40, 41, 42, 44. At wavelengths of 8 m and longer, with equipment of equivalent sensitivity, these “visual-type” echoes are usually drowned in an exceptionally numerous background, which shows a noticeable annual and diurnal variation. These variations have been studied in detail27, 37, 43. It has become clear that the number of echoes is maximal between midnight and six o’clock and minimal after noon, and that the general level of activity is higher in autumn than in spring. These variations can be explained by the motion of the Earth through meteor matter distributed in space. There is no close correlation between visible meteor showers and this abundant background, which exhibits diurnal and annual variation.

These effects are illustrated by Fig. 12, which shows the number of short-duration echoes observed simultaneously

at 4 and 8 m, with apparatus using similar antennas, over the course of 24 hours—from 9:00 a.m. GMT on September 23 to 9:00 a.m. GMT on September 24, 1947. These observations indicate two important facts:

a) there are unquestionable experimental data (see, for example, ^{40,44}) showing that the echoes indicated by the lower curve are associated with meteors of the type encountered in known visible meteor streams;

b) during these observations all the echoes observed at 4 m were also observed at 8 m, but in addition, at 8 m a very intense background with a diurnal variation was observed. Up to now there are no direct experimental data showing that these numerous echoes are associated with meteors. It is usually assumed that this is so, first, because these echoes are indistinguishable from the echoes produced by meteors for which a visual correlation can be obtained, and, second, because the annual and diurnal variations are such as should be expected on the basis of the general distribution of meteoric matter through which the Earth must pass.

The true nature of the second type of “meteor” echoes remains to be further investigated. If, as seems highly probable ^{27}, it is associated with uniformly distributed meteoric dust too fine to produce visible meteors, then the question of whether this dust is associated with the solar system or comes from interstellar space is of great astronomical interest.

B. Study of the abundance of meteors in the main meteor streams. Apart from cases of exceptionally high meteor abundance in meteor streams, the echo background described above masks the influence of ordinary meteor streams on radio apparatus operating at a wavelength of 8 m and longer. In the four-meter range the abundance of the background is very low and is associated only with meteors of weak radiants and with random single meteors. This makes it possible to study the main meteor streams in detail.

The first measurements of the main meteor streams were carried out at a wavelength of 4–5 m ^{40}. Fig. 13 shows the mean number of echoes per hour obtained during the first half of 1946. The maximum in January corresponds to the Quadrantid stream, and the maximum in April to the Lyrids, with both maxima coinciding in time with the visual maxima of these streams.

Similar curves of meteor activity during the summer of 1946, including the Perseid stream, were published by Prentice, Lovell, and Banwell ^{44}.

The remarkable Draconid meteor shower of October 10, 1946, was studied in England and in America ^{26,27,41,45,46*}. The abundance of echoes

) It was also observed by means of radar stations in the USSR ^{47}. Editors*

during the night of October 9–10, 1946, at a frequency of 27 MHz, is shown in Fig. 5. This observation is an exceptional case in which a visible-type meteor stream produced an enormous effect, exceeding the high background rate observed at these frequencies.

A similar rate curve, obtained at a frequency of 72 MHz, is shown in Fig. 14^41. The normal rate outside streams on this equipment is about two echoes per hour, and these conditions held until 0 h UT on October 10. But at about 3 h 40 min UT the echo rate reached a tremendous maximum of 168 echoes per minute, and by 6 h 00 min UT it had again fallen to the normal non-stream rate.

Fig. 13. Mean echo rate according to observations at a station with a vertical beam.

Since October 1946, the activity of the principal meteor streams has been systematically studied by Lovell and his collaborators at a frequency of 72 MHz. The case of the Draconid meteor shower is, of course, unique both in its activity and in its short duration. A more normal behavior of the principal streams is illustrated by the rate curve for the Geminid stream for December 1946, shown in Fig. 15.

B. Determination of meteor radiants. In 1938, attention was drawn^36 to the influence of the orientation of the meteor trail with respect to the observing apparatus. Further experimental investigations^40,41 showed that, indeed, in observing meteor trails there is an effect of the angle,

Fig. 14. Number of short-duration echoes from meteors of the Draconid stream on October 10, 1946, according to measurements from motion-picture images of the oscilloscope screen (frequency 72 MHz, antenna—a half-wave dipole).

Fig. 15. Average hourly number of Geminids during the night (observations at a frequency of 72 MHz).

so that a strong radio echo is obtained only when the radio beam is directed perpendicular to the trail*). This property of meteor trails has been used to determine the radiants of streams by observing the radio echo.

a. The first determination of meteor radiants by the radio method was carried out in 1947.^40 The authors used three separate stations operating at a frequency of 73 MHz and located as shown in Fig. 16. The antennas were turned so that the beams intersected at a point at an altitude of 100 km, equidistant from all the stations. Since a meteor trail gives a radio echo only if it is directed perpendicular to the axis of the beam, it was to be expected that, when the radiant moves across the sky, it would not produce an echo simultaneously for all three stations, but only for those for which the meteor trails are oriented at right angles to the radio beam**).

Fig. 16. Location of the stations operating in June–July 1945.

The diurnal variation of the mean hourly numbers obtained at stations B1, B2, B3 (Fig. 16) between July 26 and August 1, 1945, is shown in Fig. 17. The regions of possible positions of the radiant corresponding to the maximum number of echoes for B2 at 2 h 30 m and for B3 at 4 h 30 m are shown in Fig. 18. The center \(R\) of the region of overlap of these areas may be taken as the position of the radiant, and this gives the following equatorial coordinates: \(\alpha = 345^\circ,\ \delta = -10^\circ\). For station B1 no noticeable maximum was obtained. This was to be expected, since a radiant having such coordinates is never in a position

) This effect is sharply expressed at wavelengths shorter than \(6 \tfrac{1}{2}\) m. At longer wavelengths, strong reflection is obtained not only when the ionic column is irradiated at right angles to it. At long wavelengths it is also possible to observe intense reflection from the head of an approaching meteor and reflection from a receding meteor. The possibility also arises of “triangulating” a meteor trail by simultaneous observations with the aid of three stations located at large distances from one another.^49,50 Editors.*

) As already noted in the preceding note, this is true only for sufficiently short wavelengths. Editors.

favorable for \(B1\). The obtained radiant agrees well with the Delta Aquariid radiant, which, as is known, is active on these dates.

Fig. 17. Diurnal variation of the mean hourly number of echoes in the period from July 26 to August 1, 1950. Periods when the radiant \(R\) occupies a favorable position are marked with bold lines.

The accuracy of determining the radiant coordinates depends on the width of the radiation lobe. The authors believe that their error in determining the position of the radiant, caused by this reason, may amount to \(10^\circ\).

Fig. 18. Regions of possible positions of the radiant corresponding to the principal maxima of hourly numbers for stations \(B2\) and \(B3\).

1. Another method for determining meteor radiants, using a movable antenna with a narrow beam at a single station, was developed by Clegg[^51]. To determine the right ascension of the radiant, the beam is directed exactly to the east. Echoes will not be observed until the meteor trails cross the beam at a right angle, i.e., until the radiant is exactly in the south. Consideration of the geometry of the intersection of the radio beam with the layer at an altitude of \(100\) km, in which meteors produce ionization, shows that echoes appear at large distances and that, as the radiant moves,

distant to the west, the echo range continuously decreases. When the radiant emerges from the beam, the echoes disappear at the very smallest ranges. The time of the first appearance of echoes at a large range coincides with the time of culmination of the radiant, from which its right ascension can be calculated.

To determine the declination of the radiant, the beam is turned through some angle to the south, so that it again lies ahead of the radiant. When the radiant reaches a position perpendicular to this new direction of the beam, the echoes again appear at large ranges, and their ranges decrease as the radiant moves farther and farther to the west. Knowing the interval of time between the first appearances of echoes at large range in the two positions, i.e., knowing the time during which the radiant has turned through an angle $\theta$ after culmination, one can calculate the declination of the radiant.

Fig. 19. Echo ranges as a function of time. Observations on May 7, 1947, with a narrow fan-shaped beam directed eastward for the purpose of determining the positions of radiants by Clegg’s method. Between $7^{\mathrm h}\ 30^{\mathrm m}$ and $9^{\mathrm h}\ 30^{\mathrm m}$ the passage through the beam of the Eta Aquarid radiant was observed (the dotted line shows the theoretical range curve). The subsequent echoes belong to a new daytime stream, whose passage begins at $10^{\mathrm h}\ 35^{\mathrm m}$ (a stream from the constellation Pisces).

Fig. 19 shows the character of the results obtained by this method. It gives the ranges of individual echoes observed on May 7, 1947, with the beam directed exactly eastward. The first stream was the Eta Aquarids, which culminated at $7\ \mathrm h.\ 40\ \mathrm m.$ GMT. When this stream passed through the beam, a second radiant appeared at $10\ \mathrm h.\ 40\ \mathrm m.$ GMT, which disappeared at about $12\ \mathrm h.\ 30\ \mathrm m.$ GMT. Subsequent counts at another azimuth made it possible to determine its declination, which showed that this second stream was from a radiant in the constellation Pisces (see Section 7, G).

The accuracy of this method depends on the horizontal width of the beam, on the number of echoes, and on the declination of the radiant. With a given

for a given beam width this method gives a considerably greater accuracy than the method described in the preceding section, since it uses the measured ranges of echoes. The apparatus used makes it possible to determine the position of the mean radiant producing echoes at a rate of about 30 per hour, with an accuracy of \(\pm 3^\circ\) in the direction perpendicular to the celestial equator, and \(\pm 1^\circ\) in the direction parallel to it.

G. Discovery of daytime meteor streams. The methods described above make it possible to study meteor streams by day, and the preliminary results of daytime observations reveal some surprising facts. First, a meteor stream was found, active between June 6 and 13, 1945, with coordinates \(\alpha = 58^\circ\), \(\delta = +5^\circ\), and measurements during the summer of 1946[^44] revealed the presence of considerable meteor activity in the daytime. A systematic study of the principal streams, begun in October 1946, showed that in autumn and winter the general predictions of the activity of meteor streams, based on visual observations, are in the main correct. But in the investigation of the \(\eta\)-Aquarid radiant, which began on May 1, 1947, it was found that the meteor stream observed by astronomers is only the beginning of an exceptionally active belt extending in the direction of the Sun.

The positions of the radiants were determined by Clegg’s method, and it turned out that the main radiant lies in Pisces. Since its culmination occurs between 9 and 11 a.m., the study of this stream by means of visual observations is impossible.

Observations*) carried out in 1947 and 1948 made it possible to study daytime streams active in the summer months[^52]. Nine separate streams were identified, some of which have several radiants (or a large radiant area), situated in the constellations Pisces, Perseus, Aries, Taurus, Orion, and Gemini. For some of these streams, at maximum the number of meteors reaches 80–100 per hour.

During August no conspicuous daytime streams are observed, but in June–July there are radiants (culminating early in the morning) which for the present remain unstudied.

In 1946 and 1947 observations were conducted in mid-November and at the beginning of December for the purpose of studying very weak streams associated with Biela’s comet[^53]. The observations yielded few results because of the small number of meteors. At the end of December 1947 observations were made of the Ursid stream, recently discovered by Bečvář. It was possible to determine the position of the radiant and the population of the stream[^54].

D. Determination of meteor velocities. During the Draconid star shower in 1946 it was possible for the first time to carry out

*) Editor’s addition (to the end of the section).

determination of meteor velocities from radio-echo observations[^45]. Instead of the usual amplitude sweep, in which the signals produce deflections on the linear range sweep on the screen of a cathode-ray oscilloscope, the incoming echoes modulated the brightness of the linear range sweep. This sweep was photographed on motion-picture film moving in a direction perpendicular to it at a speed of 2.4 mm per second.

In this way a continuous photographic record was obtained of the ranges of the echoes, the moments of their appearance, and their durations.

Fig. 20. Photographic registration of a short-duration ionospheric echo during the Draconid meteor shower in 1946.

In many of these photographs remarkable rapidly moving echoes are noticeable, preceding the formation of the main echo. An example of such an echo is shown in Fig. 20, in which one can see a faint trace beginning at a point with approximate coordinates 103 km, 1.75 seconds and moving toward the main echo, which extends approximately from 2.75 to 5 seconds at a constant range of about 100 km.

The authors attributed this faint trace to energy scattered from the head of the ion column produced by the approaching meteor, whereas the main intense echo is an ordinary reflection arising when the meteor trail approaches the foot of the perpendicular dropped from the observing station onto the trajecto-

meteor. Thus, a weak trail gives a relation between the variable range and the time for the approaching meteor. If the meteor has velocity \(V\), then the range \(R\) at the moment \(T\) is given by the formula

\[ R^2 = R_0^2 + V^2 (T - T_0)^2, \]

where \(T_0\) is the moment when the meteor reaches the minimum range \(R_0\). The velocity was determined from three observed values of \(R\) and \(T\) by eliminating \(R_0\) and \(T_0\). By this method the velocity was computed for 22 Draconid meteors. On the average, \(22.9 \pm 1.3\) km/sec was obtained, in excellent agreement with the theoretical geocentric velocity \(23.7\) km/sec.

Later*) it became possible to determine meteor velocities from diffraction during the formation of the trail\(^{55}\). Changes in the amplitude of the echo, occurring while the meteor is still moving, are associated with the fact that diffraction maxima and minima pass through the observing station; these can be calculated with the aid of Fresnel integrals.

To record the changes in amplitude, an additional recorder was used, in which a horizontal sweep of duration 0.2 second was triggered by the reflected signal. On this recorder the amplitude of each echo (for each transmitted pulse) was photographed during 0.2 second from the moment of appearance of the meteor trail.

At first this method was applied to determining the velocities of the Geminid (1947) and Quadrantid streams. For the Geminids the velocity obtained was \(34.4 \pm 2.8\) km/sec, in excellent agreement with the result of Whipple’s photographic observations, \(34.7\) km/sec. For the Quadrantids the velocities fell into three groups, and velocity determinations by other methods are lacking.

In 1948 this method was used to study the velocities of meteors of daytime streams\(^{56}\), discovered with the aid of radar observations. It proved possible to determine the geocentric velocities of two streams, which, after recalculation into heliocentric velocities, showed that these streams possess small elliptical orbits.

For the determination of velocities by recording meteor whistles, see p. 41.

8. Application to the Physics of the Upper Layers of the Atmosphere

A. Intensity of scattering of radio waves. a. Experiments

Eckersley’s\(^{7,37}\) observations of short-lived echoes were carried out at wavelengths exceeding 15 m, i.e., such that

*) Editorial addition (to the end of the paragraph)

The echo background, which shows daily and annual variations, dominates over the effect of meteor streams. The author believed that the echoes are produced by ion clouds existing for more than 0.02 seconds but less than 1.0 second, and considered the scattering of radio waves by such clouds to be analogous to the scattering of α-particles by heavy atoms. According to his theory, the intensity of backscattering should be proportional to \(\lambda^4\).

Fig. 21. Scattering intensity as a function of wavelength according to observations from various stations, reduced to a standard power of 5 kW. For comparison the fourth-power law is shown.

In Fig. 21 a comparison is given of experimental data for scattering intensity over the wavelength range from 15 to 30 m with the predicted fourth-power law.

1. Pierce\(^35\) calculated the ratio of the energy scattered by the meteor trail to the energy scattered by an ionized layer, under the assumption that in the trail the electrons form a long column whose diameter is large in comparison with the wavelength.

In Fig. 22 \(AB\) is the equivalent reflecting plane of the ionized layer, and the circle with center at \(C\) is the equivalent reflecting cylinder whose axis coincides with the meteor trajectory. The energy emitted from \(O\) and reflected from \(AB\) in the direction toward \(O\), and falling on a small surface element at \(O\), is

\[ E_p=\frac{k}{h^2}. \]

For the cylinder at \(C\), if the density in the direction perpendicular to the axis of the cylinder is proportional to \(\dfrac{r}{d}\) and is homogeneous along the axis, then the reflected energy is equal to:

\[ E_c=\frac{k}{d^2}\,\frac{r}{d}=\frac{kr}{d^3}. \]

Thus,

\[ \frac{E_c}{E_p}=\frac{rh^2}{d^3}. \]

The author gives, as an example, the case of reflection from a meteor trail at a distance \(d=100\) km with \(r=0.25\) km, in comparison with reflection from a layer \(F\) at a height of 200 km, and obtains:

Fig. 22.

\[ \frac{E_c}{E_p}=0.0156. \]

The ratio of the corresponding field strengths is equal to:

\[ \sqrt{\frac{E_c}{E_p}}=12.5:100. \]

The author mentions that this ratio agrees satisfactorily with the experimental values of field strength found by him in “bursts” (see Section 5), but gives no more detailed experimental data that could confirm the correctness of his approach to the problem. According to Herlofson\({}^{13}\), these calculations lead to densities in the meteor trail that may be in error by many times.

c. The suggestion that some of the short-duration radio echoes may be caused by coherent scattering by electrons in a long column whose diameter is small compared with the wavelength \(\lambda\) was made by Blackett and Lovell\({}^{57}\). This idea was applied to the special case of meteor trails\({}^{58}\). In this case the number of electrons \(N\) scattering coherently is equal to their number in a section of the column whose length is equal to the first Fresnel zone, i.e.,

\[ N=\alpha\sqrt{\frac{\lambda R}{2}}, \]

where \(\alpha\) is the number of electrons produced by the meteor per centimeter of path.

Introducing the parameters of the radio apparatus, it can be shown that

\[ \alpha=\sqrt{24\pi\,\frac{mc^3}{e^2}\,\frac{1}{G}}\sqrt{\frac{\varepsilon R^3}{P\lambda^3}}, \tag{1} \]

where \(G\) is the gain factor of the receiving and transmitting antennas relative to a half-wave dipole, \(\varepsilon\) is the scattered power returning to the receiver, \(R\) is the distance to the trail, \(P\) is the peak power of the transmitter, and \(\dfrac{mc^3}{e^2}\) is the reciprocal of the classical electron radius.

The predicted dependence between \(\varepsilon\) and \(\lambda\) was tested experimentally \(^{58}\) by simultaneous observations of radio echoes from one and the same meteor trail at different wavelengths. Within the limits of observational error, the dependence agrees with that predicted over the wavelength range used, namely from 1.4 to 8 m. The measured values of \(\alpha\) for visually observed meteors, based on this formula, also agree well with theory \(^{18}\).

c. The available observational data are still insufficient to decide which of the interpretations given above is correct. Lovell’s interpretation \(^{58}\) agrees with the experimental data with respect to the dependence of \(\varepsilon\) on \(\lambda\) in the investigated wavelength range from 1.4 to 8 m and gives an electron density that agrees well with theory. It also correctly predicts the change in the number of registered echoes when \(\lambda\) changes from 1.4 to 6 m. However, it is wholly unable to explain the sharp increase in the number of echoes at 8 m as compared with 6 m, described above in section 7,A.

The formula \(\sim \lambda^4\) agrees with experiment in the range from 15 to 30 m, but it is inapplicable to wavelengths shorter than 8 m, where shower meteors dominate. This formula, like Peirce’s calculation \(^{35}\), gives electron densities that disagree with modern theoretical estimates \(^{13}\).

It seems possible that the background of high abundance at long wavelengths is caused by meteoric dust, which creates electron columns whose properties differ from those created by shower meteors. Indeed, it is difficult to imagine that the properties of both types of ionization would be the same, since, as indicated above, in observations at 8 m the “invisible background meteors” give echoes of the same character and amplitude as those associated with visually observed meteors.

B. Electron density in meteor trails. By carrying out simultaneous visual observations and radio-echo observations, one can measure the amplitude of the radio echo produced by a visible meteor. Such observations were carried out for the Perseids of 1946 \(^{44}\), the Draconids of 1946 \(^{41}\), and the Perseids of 1947 \(^{41}\).

Using formula (1), it was possible to calculate \(\alpha\)—the number of electrons produced per centimeter of path—since \(G\), \(P\), \(\lambda\)—

known apparatus constants, while \(\varepsilon\) and \(R\) are measured for each trail.

Three series of observations agree well with one another and give a value of \(\alpha\) between \(10^9\) and \(10^{12}\) electrons per centimeter of path.

Detailed comparisons of \(\alpha\) with visual stellar magnitudes have not yet been made; however, it is evident that meteors of approximately the 5th stellar magnitude, close to the limit of visibility with the unaided eye, produce about \(2 \cdot 10^{10}\) electrons per centimeter of path. This agrees well with theoretical estimates\(^{13}\). It is interesting to note that, if these estimates are correct, the visible meteor has a radius of approximately \(1/10\) mm and a mass of several milligrams.

Pierce\(^{25}\) calculated the energy brought into the atmosphere by the entire Draconid meteor shower in October 1946 and obtained an energy flux of \(3\) W per km\(^2\). On the basis of Mersis’s theory of meteor ionization\(^{12}\), Pierce calculated the rate of ion formation and obtained \(42\) ions per cm\(^3\) per second. According to Herlofson\(^{13}\), these estimates may be wrong by many times*).

B. Durations of short-lived echoes.
The distribution of the durations of short-lived echoes has been studied by many investigators\(^{29,37,40,44}\), and all of them have consistently found that the distribution has the form shown in Fig. 10.

Recent investigations, in which one and the same meteor trail was observed simultaneously at two wavelengths (unpublished results of Lovell and his collaborators), confirmed the earlier assumption that the echo duration increases with increasing wavelength. Thus, although the general character of the distribution curve remains the same, the lower limit shifts toward shorter durations as the frequency increases. Measurements at a frequency of 27 MHz established\(^{27}\) that durations after midnight prove to be longer than before midnight. These results are interpreted by the authors as an indication that in the second half of the night stronger reflections are obtained than in the first half.

So far there is no satisfactory explanation of what factors determine the duration of short-lived echoes. It is clear only that the height of formation of the trail in the atmosphere is not the sole significant factor. Although the heights of individual echoes have not yet been measured accurately by any of the investigators, all observations indicate that echoes in the height interval \(100 \pm 5\) km can have durations differing from one another by as much as a thousandfold.

) Estimates of the electron density are also contained in \(^{59}\). Editors.*

Further, a systematic study of the principal showers showed that the distribution of echo durations for the principal showers does not reveal any appreciable differences. Fig. 23 shows the distribution of durations for the Geminid, Leonid, Orionid, and Draconid showers of 1946 and the Quadrantids of 1947. The velocities of these meteor showers range from 23.7 km/sec (Draconids) to 72 km/sec (Leonids), and visual observations reveal

Fig. 23. Distribution of echo durations for five principal showers.

a clear relation between the height at which the meteor appears and its velocity. Meteors of low velocity become noticeable at a height of approximately 95 km, while fast meteors at a height of about 120 km. The atmospheric pressure changes by a factor of 10 within this layer, and if the processes of recombination and attachment were the only ones determining the echo duration, then this pressure effect would be noticeable in Fig. 23. The similarity of the curves thus shows that the rate of recombination and attachment, determined by the height at which the meteor appears in the atmosphere, is not the sole factor determining the duration of the radio echo.

It has been suggested¹³ that the principal factor determining the echo duration may be not recombination, but electron diffusion. Depending on the direction of the meteor trail

with respect to the lines of force of the Earth’s magnetic field, diffusion will occur in such a way that the electron column will either expand or lengthen. Thus, the duration of the radio echo should depend on the direction of the antenna—on whether the echo is observed from a column that is lengthening, or from a column that is expanding. In the latter case the interference effects will rapidly reduce the intensity of scattering, and the echo will last for a shorter interval of time than in the former case.

The existing experimental data agree well with this theory: a) if diffusion determines the duration, then, for a given sensitivity of the apparatus, the duration of the echo must depend on the wavelength. Simultaneous observations at two wavelengths of echoes from the same meteor trails show that the duration is proportional approximately to \(\lambda^{2}\); b) if the direction of the trail with respect to the Earth’s magnetic field is important, then the duration of the echo from meteors of one and the same stream will depend on the direction of the radio beam.

Fig. 24. Mean echo durations as a function of the direction of observations.

In investigations carried out at the experimental station of the University of Manchester (\(53^\circ 13'54''\) N lat.; \(2^\circ 18'11''\) W long.), using a directional antenna, the mean duration of the echo was measured as a function of the antenna azimuth (in the northern direction) for the Leonids and Geminids of 1946 and for the stream from the constellation Pisces in 1947. In all cases maxima of the mean echo duration were obtained between \(0^\circ\) and \(20^\circ\) east longitude. The results for the Geminids and for the stream from the constellation Pisces are shown in Fig. 24, in which the mean echo durations are plotted as a function of the antenna azimuth.

G. Echo fluctuations and motions and possible applications to the study of winds

Radio echoes obtained from meteor trails exhibit, over the course of their lifetime, very noticeable fluctuations in intensity. The technique for studying them by means of motion-picture photography for echoes of long duration, and examples of fluctuations obtained in this way, were described by Lovell, Banwell, and Clegg[^41]. Further (unpublished) investigations, based on motion-picture records of several hundred meteor echoes, showed that almost all echoes exhibit significant amplitude fluctuations superposed on the decay after the initial maximum. These fluctuations are usually chaotic in character when successive photographs taken at intervals of \(1/8\) second are compared, but sometimes they are grouped into secondary maxima

Fig. 25. Fluctuations of echo amplitude according to measurements at a frequency of 62 MHz
(meteor from a radiant in the constellation Pisces).

with durations of about a second or more and separated by intervals of several seconds.

An example of an echo of this type is shown in Fig. 25, in which the echo amplitude is given in comparison with the height of the noise level, and the time in seconds is counted from an arbitrary zero chosen near the moment of the beginning of the echo. Such changes may be associated with distortion of the ionized column by stratospheric winds[^13]. It has also been established that very rapid fluctuations, with periods on the order of \(1/10\) second or less, are superposed on the longer-period fluctuations mentioned above. There is as yet no explanation of this phenomenon. It has been found[^40] that about \(2\%\) of all observed echoes exhibit a displacement in range, and among echoes lasting more than 1 second such displacement is found in \(15\%\) of echoes. Echoes lasting more than 0.5 second have radial velocities of these displacements usually lying between 0 and \(-5\) km/sec. Only very few have higher velocities—from \(+5\) to \(-25\) km/sec. Most of these motions must signify a phenomenon different from the ionization sometimes observed in the head of a moving meteor, which has been used to determine velocities. They are probably connected with the drift of the ionized column under the influence of stratospheric winds. It is possible that further investi-

...observations of fluctuations and drift of echoes can provide information about winds at great altitudes.

D. Distribution of meteor masses. As described above (pp. 35–36), observations of short-duration echoes make it possible to calculate the electron density in the meteor trail. Assuming that formula (1) is valid for shower meteors, the distribution of electron densities was calculated^41 for meteors of the Draconid shower of 1946. This was done from measurements of the amplitudes and ranges of echoes on a motion picture film taken near the maximum of the shower with apparatus operating at a frequency of 72 Mc/s. It was found that the distribution of densities obeys an inverse proportionality law with an exponent equal to \(0.9 \pm 0.2\). Further measurements for the Geminids of 1946 gave an exponent of \(0.9 \pm 0.2\), and measurements for the Quadrantids of 1947 gave \(1.0 \pm 0.2\). Thus, the distribution of electron densities among shower meteors agrees with an inverse proportionality law, with an exponent close to unity, over the interval of electron densities covered by the measurements (from \(10^9\) to \(10^{12}\) electrons per centimeter of path).

If it is assumed that the velocities of the meteors of a given shower are the same, then it follows from the theory of meteor ionization that the mass distribution of the meteors must be the same as the distribution of electron densities. Consequently, the measurements described above indicate that the number of meteors of a given mass in these showers is inversely proportional to the mass of the meteors.

This result is consistent with the existing visual observations, which indicate that for each stellar magnitude the total mass of meteoric matter is the same. Further study of the mass distribution by means of radio-echo observations is of especially great interest if it can be applied to meteors fainter than the limit of visibility.

1. The phenomenon of “whistling meteors.” In 1941 attention was drawn^60 to faint whistles of an unusual character when tuning to an unmodulated carrier wave of frequency 7 Mc/s at a distance of 10 miles from the Delhi short-wave transmitter. These whistles were perceived as high-pitched sounds, the pitch rapidly decreasing. They were of short duration—usually from \(1/5\) second to several seconds—and occurred at irregular intervals. By establishing visual correlations it was shown that these whistles were associated with the flight of meteors. According to these investigators, the explanation of the whistles is that they are beat sounds arising as a result of interference between the radio wave reflected from the head of the ion column produced by the approaching meteor and the direct ground wave. To explain the falling pitch it was proposed that the velocity of the meteor decreases...

as it plunges into the atmosphere. In this case the scattered wave, interfering with the ground ray, gives a whistle of gradually decreasing pitch.

These experiments were analyzed^61, and doubt was expressed about the explanation of the whistles given above; it was suggested that some of them may be caused by atmospherics, and some by meteors creating an “electric impulse” when they are detained by the atmosphere. However, the connection of whistles with meteors received further confirmation. During the 1946 Draconid shower, listening was carried out on two types of apparatus^6: a) 72 miles from a 100-kilowatt transmitter at 15 MHz, and b) 1 mile from a 0.7-kilowatt transmitter at 29 MHz. On both receivers Doppler whistles were observed, coinciding with meteors observed visually, and, often, the whistles were accompanied by flashes of a strong signal.

Pierce^26 also observed Doppler whistles during the Draconid shower and came to the conclusion that the explanation according to which the whistles are connected with the braking of the meteor is incorrect; he assumes that the Doppler beating is created by an “ionic head wave” and is a function both of the meteor velocity and of the diffusion velocity of the electrons.

Appleton and Naismith^27 also believe that the explanation connected with the braking of the meteor is incorrect; the pitch of the whistle should decrease also in the case when the wave is reflected from the head of a column approaching at constant speed. When the meteor reaches the foot of the perpendicular dropped from the station to the meteor trajectory, the pitch should fall to zero. But this is the point at which the meteor trail creates the main radio echo, and, consequently, the whistle must always precede the echo if this explanation is correct. Observations^62, according to which flashes of intense signal followed the whistles, speak in favor of this explanation.

Investigations of meteor velocities indicate that this explanation is almost certainly correct. No noticeable braking of the meteor has been found, but photographic records have been obtained of a rapidly moving echo caused by ionization in the head of an approaching meteor.

In subsequent works*) on the study of meteor whistles, formulas are given for determining the true paths of meteors in the atmosphere and the positions of radiants from the curves of whistle pitch height obtained at three stations^63, and the results are presented of determining meteor velocities^64 by recording the whistle pitch at frequencies of 12 and 30 MHz, which are in satisfactory agreement with astronomical data.

*) Editorial addition (to the end of the paragraph).

References Cited

1. H. R. Mimnio. Rev. Mod. Physic 2, 1 (1937).
2. Ya. L. Alpert, Propagation of Radio Waves in the Ionosphere, Gostekhizdat, 1947.
3. M. Eigeneson, M. I. Gnevyshev, A. I. Ol, B. M. Rubashev, Solar Activity and Its Terrestrial Manifestations, Gostekhizdat, 1948.
4. V. L. Ginzburg, Theory of the Propagation of Radio Waves in the Ionosphere, Gostekhizdat, 1949.
5. Ya. L. Alpert, UFN 34, 262 (1948); 36, 1 (1948).
6. R. A. Heising, Proc. Inst. Radio Eng. 16, 75 (1928).
7. T. L. Eckersley, J. Inst. Electr. Engr. 67, 992 (1929); 71, 405 (1932).
8. E. V. Appleton and R. Naismith, Proc. Roy. Soc. A137, 36 (1932).
9. J. P. Schafer and W. M. Goodall, Proc. Inst. Radio Eng. 19, 1434 (1931).
10. H. Nagaoka, Imp. Acad. Tokyo Proc. 5, 233 (1929); Inst. Phys. Chem. Res. Sci. Papers, Tokyo 15, 169 (1931).
11. A. M. Skellett, Phys. Rev. 37, 1668 (1931); Proc. Inst. Radio Eng. 20, 1933 (1932); 23, 132 (1935).
    11a. H. A. Ivanov, Astr. Zhurn. 8, 240 (1931).
12. H. B. Maris, Terr. Magn. Atm. Electr. 34, 309 (1929).
13. N. Heerlofson, Rep. Progr. Phys. 11, 444 (1948).
14. T. Minohara and Y. Ito, Rep. Radio Res. Japan 3, 115 (1933).
15. E. Quäck, Elektr. Nachr.-Techn. 8, 46 (1931).
16. G. W. Pickard, Proc. Inst. Radio Eng. 19, 1166 (1931).
17. I. Ranzi, Nature 130, 368 (1932).
18. E. V. Appleton and R. Naismith, Proc. Phys. Soc. 45, 389 (1933).
19. J. A. Ratcliffe and E. L. C. White, Proc. Phys. Soc. 45, 399 (1933); 46, 107 (1934).
20. L. R. Hafstad and M. A. Tuve, Terr. Magn. Atmos. Electr. 34, 39 (1929).
21. E. V. Appleton, R. Naismith and L. J. Ingram, Phil. Trans. Roy. Soc. A236, 191 (1937).
22. E. V. Appleton and J. S. Hey, Phil. Mag. 37, 73 (1946).
23. M. N. Gnevyshev, Climate and Weather 10, No. 2, 17 (1934).
24. S. K. Mitra, P. Syam and B. N. Ghose, Nature 133, 533 (1934).
25. J. N. Bhar, Nature 139, 470 (1937); Indian J. Phys. 11, 109 (1937).
26. J. A. Pierce, Phys. Rev. 71, 88 (1947).
27. E. V. Appleton and R. Naismith, Proc. Phys. Soc. 59, 461 (1947); Nature 158, 936 (1946).
28. J. A. Ratcliffe and J. L. Pawsey, Proc. Cambr. Phil. Soc. 29, 301 (1933).
29. A. M. Skellett, Nature 141, 472 (1938).
30. R. A. Watson-Watt, A. F. Wilkins and E. G. Bowen, Proc. Roy. Soc. A161, 181 (1937).
31. E. V. Appleton and J. H. Piddington, Proc. Roy. Soc. A164, 467 (1938).
32. T. L. Eckersley and F. T. Farmer, Proc. Roy. Soc. A184, 196 (1945).
33. L. Harang, Geofysiske Publikas (Oslo) 13, No. 4 (1942).
34. J. A. Pierce, Phys. Rev. 59, 625 (1941).
35. J. A. Pierce, Proc. Inst. Radio Eng. 26, 892 (1938).
36. Federal Communications Commission, Engineering Dept., Docket No. 6651 (1944).
37. T. L. Eckersley, J. Inst. Electr. Eng. 86, 548 (1940).
38. H. W. Wells, J. M. Watts and D. E. George, Phys. Rev. 69, 540 (1946).

1. A. C. B. Lovell, J. A. Clegg and C. D. Ellyett, Nature 160, 372 (1947).
2. J. S. Hey and G. S. Stewart, Proc. Phys. Soc. 59, 858 (1947).
3. A. C. B. Lovell, C. J. Banwell and J. A. Clegg, Mon. Not. R. Astr. Soc. 107, 164 (1947).
4. O. P. Ferrell, Phys. Rev. 69, 32 (1946).
5. E. Eastwood and K. A. Mercer, Proc. Phys. Soc. (1948).
6. J. P. M. Prentice, A. C. B. Lovell and C. J. Banwell, Mon. Not. R. Astr. Soc. 107, 155 (1947).
7. J. S. Hey, S. J. Parsons and G. S. Stewart, Mon. Not. R. Astr. Soc. 107, 176 (1947).
8. J. Q. Stewart, M. Ference, J. J. Slattery and H. A. Zahl, Sky and Telescope 6, 3 (1947).
9. B. Yu. Levin, Astr. Circ. No. 57, 2 (1946).
10. A. C. B. Lovell, Nature 161, 279 (1948); P. M. Millman, McKinley and M. S. Burland, Nature 161, 278 (1948).
11. P. M. Millman and McKinley, J. R., Astr. Soc. Canada 42, 121 (1948).
12. P. M. Millman and McKinley, Sky and Telescope 8, 114 (1949).
13. J. A. Clegg, J. Brit. Astr. Assoc. 58, 271 (1948); Phil. Mag. 39, 577 (1948).
14. A. Aspinall, J. A. Clegg and A. C. B. Lovell, Mon. Not. R. Astr. Soc. 109, 352 (1949).
15. A. C. B. Lovell and J. P. M. Prentice, J. Brit. Astr. Assoc. 58, 140 (1948).
16. J. A. Clegg, V. A. Hughes and A. C. B. Lovell, J. Brit. Astr. Assoc. 58, 134 (1948).
17. C. D. Davis and C. D. Ellyett, Nature 161, 596 (1948); Phil. Mag. 40, 614 (1949).
18. C. D. Ellyett, Mon. Not. R. Astr. Soc. 109, 359 (1949).
19. P. M. S. Blackett and A. C. B. Lovell, Proc. Roy. Soc. A177, 183 (1941).
20. A. C. B. Lovell and J. A. Clegg, Proc. Phys. Soc. 60, 491 (1948).
21. E. W. Allen, Proc. Inst. Radio Eng. 36, 346 (1948).
22. Chamanlal and K. Venkataraman, Electrotechnics 14, 28 (1941).
23. S. R. Khastgir, Indian J. Phys. 17, 239 (1943).
24. L. A. Manning, R. A. Helliwell, O. C. Villard and W. E. Evans, Phys. Rev. 70, 767 (1946).
25. D. D. Cherry and C. S. Shyman, Phys. Rev. 75, 1441 (1949).
26. D. W. Heightman and T. W. Bennington, Wireless World 53, 219 (1947).
27. L. A. Manning, J. Appl. Phys., No. 8 (1948).
28. J. A. Clegg, V. A. Hughes and A. C. B. Lovell, Mon. Not. R. Astr. Soc. 107, 369 (1947).
29. M. Hurohata, Publ. Astr. Soc. Japan 1, 39 (1949).
30. A. C. B. Lovell, J. Inst. Electr. Eng. 95, 324 (1948).