zotero/storage/YBJKFRXL/.zotero-ft-cache

VOI.UMr. 9

JANUARY, 1937

NUMazR I

The Physics of the Ionosyhere
HARRY ROWE M IMNO IIarvard University, Cambridge, Massachusetts

TABLE OF CONTENTS

A. Introduction. . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . 1

B. Brief historical survey. . . . . . . . . . . . . . . . . . . . . . . . . 2

C. Basic exyerimentai facts. . . . . . . . . . .

5

D. Elementary theory. . . . . . . . . . . . . . . . . . . . . . . . . . . 11

E. The nature of the accelerating field. . .

14

F. Other forces acting on the electron. . . . . , . . . . . . . 15

G. The equations of moti, on of the electrons. . . . . , . . 18

H. Anetysis of ~eyxeto-ionic d.ouble refraction. . . . . . 21

I. Collisional friction. . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

J. Comylete ~~&ysis by conforms& representation. . . 26

K. "Fine structure" of the ionosyhere, . . . . . . . . . . . . 27

L. Why does strati8cation existed . .

. . . . 30

M. Tidal effects in the ionosyhere. . . . . . . . . . . . . . . . . 31

N. Sunsyots, xnagnetic indices, and auroral disylays . 32

O. Magnetic storms and meteoric showers. . . . . . . . . 36

P. Thunderstorms and barometric effects. . . . . . . . . . 37

Q. Local ionosyheric clouds. . . . . . . . . . . . . . . . . . . . . . . 38

R. Scattering of radio vraves. . . . . . . . . . . . . . . . . . . . . . 40

S. Interaction of radio waves. . . . . . . . . . . . . . . . . . . . . 40

T. Ecliyse observations. . . . . . . . . . . . . . . . . . . . . . . , . . 41

U. Conclusion

. . . . . . . . . . . . . . . . . . . . . . . . . . 43

A. INTRODUCTION
~~~N October 22, 1924, an eminent British engineer delivered a lecture on "Unsolved
Problems of Wireless" before the Radio Society
of Great Britain. In his address, R. H. Bar6eld
proposed the following problems:
1. Why is long distance wireless transmission possible at all?
2. Why are signals stronger by night than by day?
3. Why do direction 6nding stations experi-
ence large errors by night, while the errors by day are practically negligible?
4. Why does the phenomenon known as fading
occur?
During the next decade, these questions and a number of allied problems were completely solved from the engineering point of view. During the course of the investigation some
additional engineering problems were disclosed,

which still await solution. These will be described
below in their proper setting. However, the most interesting results, derived
from intensive experimental research during the past twelve years, belong properly in the domain of pure physics. New and interesting experimental and theoretical problems confront the physicist. The development of new experimental tools and methods permits exact measurements in a 6eld of atmospheric physics which is not too complex for useful theoretical analysis. The theory is not complete, and the experimental
methods require extension, but a satisfactory beginning has been made in a young and active branch of physical research. The new ionosphere problems are intimately connected with a number
of other geophysical, lunar and solar investigations which have attracted the attention of
physicists for a number of years. Such related
experiments include investigations of cosmic rays, terrestrial magnetism, solar activity, auroral dis-

HARRY ROWE M I M NO

plays, igonc distribution, thunderstorm activity, atmosphere temperatures, luminous clouds, meteor trails, air-mass movements, earth currents, stratosphere meteorology, and elastic deformations of the earth's crust.
After a brief historical survey of the early radio theories and experiments, I shall attempt to present a unihed account of a number of researches carried on during the last twelve years. As a complete collection of the original publications of this active period would 611 a number of large volumes, it is necessary for me to exclude many valuable papers from this review. Therefore the papers mentioned in the bibliography have been selected from a much larger group, as they appear to illustrate adequately various points in the presentation. The references are intended to be representative, though by no means complete. Supplementary references may be found in nearly all of the papers which have been quoted. I have not adhered to a strict chronological sequence, and have not attempted to solve innumerable problems of priority.
In every active branch of' physics numerous highly controversial questions arise, and a number of them may be found in the ionosphere
6eld. When discussing such matter I believe
that the reviewer should not straddle the argument, but should formulate an unbiased personal opinion whenever sufticient evidence appears to
be available. While attempting to do this, I
have tried to avoid arbitrary statements, and have included representative references covering both sides of the dispute. The field is new and extensive and no progress can be expected without occasional justifiable errors in mathematical theory, experimental technique, and interpretation. I shall welcome correction and reproof.
When Marconi first transmitted radio signals
across the Atlantic, on December 12, 1.901, Lord Rayleigh pointed out that plane wave transmission could not account for the observed facts in a simple manner, and suggested that some sort of diffraction theory must be employed in order to explain how the wave followed the curvature of the earth.

' The diffraction theory

Rayleigh, '
Sommerfeld,

MacDonald, ' ~ Nicholson, '

was investigated by

PMoianrccahr,e,'

Zenneck, 4 von Ryb-

czynski Love, Van der Pol ' and numerous

others. In the twenty-fourth Kelvin lecture Sir

Frank E. Smith" summarized these investiga-

tions succinctly: "Many years' work by some of

the most distinguished of the world's mathe-

maticians did not sufhce to bring this apparently

innocent problem beyond a stage at which

Nicholson could say that it was one in the whole

" 6eld of mathematical analysis about which most
divergent views are held. However it is gener-

ally conceded that Watson" presented an ade-

quate survey of the problem in l9j.9, and exposed

the errors which were responsible for such wide

disagreement. The diffraction theory fails by an

enormous factor to account for the 6eld inten-

sities observed at points far below the optical

horizon. As pointed out by Larmor" in 1924, a

100-meter wave on the earth corresponds in

scale to a wave of visible light on a sphere of

6 cm radius, and it is not natural to expect a

sensible bending of the ray as a diffraction

effect alone. The elaborate mathematical analysis

merely confirms this expectation. Some divergent

views have been expressed in comparatively

recent years by Meissner" and by Kiebitz" but

the computations offered by Kiebitz have been

attacked by Mesny" with apparent success.

Since 1924, interest in the diffraction problem

has declined to a considerable extent, since

' Lord Rayleigh, Proc. Roy. Soc. 72, 40 (1904).

~ H. M. MacDonald, Proc. Roy. Soc. 71, 251 (1903);

72, 59 {1904);Roy. Soc., Phil. Trans. A210, 113 {1911);

Pro' cH. .RPooyin. cSaroec,.

A90, 50 (1914). Proc. Roy. Soc.

72,

42

{1904};Rendiconti

Circolo Matematico Palermo 29, 169 (1910); Comptes

'rendus 154, 795 (1912).
J, Zenneck, Ann. d. Physik

23, 846 (1907); Jehrbuch

der drahILosen Telegraphic, 2te. AuS. (Stuttgart, 1913);

Fir''eAJle..sWsSo.mTeNmleiecgrhrfaoeplldsho,yn, ,

tr. by Seelig (New York, 1915). Ann. d. Physik 28, 665 (1909}.
Phil. Mag. 19, 516 and 757 (1910);

Phil. Mag. 20, 157, 1910;Phil. Mag. 21, 62 and 281 (1911).

~ H. W. March, Ann. d. Physik 37, 29 (1912).

''oAWB...EvVo.anHn .RdLeyorbvcPez,yonl,RskoPiy,h.il.SAoMncn.a.,gPd.h.3ilP8.h,Ty3sr6iak5ns{4. 1129,111599, 1)1.0(519(1139)1.5). ii F E Smith, J. Inst. Elec. Eng. 73, 574 (1933);

wir"elGess.

section 9, N. Watson,

22 (1934}. Proc. Roy.

Soc. N,

83

{1919).

is J Larmor, Phil. Mag. 48, 1025 (1924).

'4 A. Meissner, Jahrbgck der drahfloserI, Telegraphic 24,

85 (1924).
"F.Kiebitz, E. N. T. 3, 376 {1926). "R. Mesny, Onde flee. 5, 650 (1926); Onde Slee. 6,

127 (1927),

PH YSI CS OF THE I Q NOSPH ERE

numerous new experimental tests have proven
conclusively that 1ong-distance transmission is mainly dependent on entirely diff'erent factors. However, diffraction undoubtedly does play an important role in certain types of short-distance transmission, and I shall have occasion to refer to the matter again in Section C.
%bile the di6raction problem was being
studied mathematically, various investigators mere also considering the several possible methods of propagation in the earth's atmos-
phere. In 1878 Stemart" had suggested that conducting layers in the upper atmosphere might account for certain types of cyclic variations in terrestrial magnetism. This hypothesis was developed further by Schuster" in 1889. In March, 1902, Kennelly" published a short article, suggesting that "There is well-known evidence that the waves of wireless telegraphy, propagated through the ether and atmosphere over the
" surface of the ocean, are reAected by that
electrically conducting surface. Three months later Heaviside" mrote a similar brief comment, which was published in December: "There may possibles be a sufficiently conducting layer in the upper air. If so, the waves will, so to speak, catch on it more or less. Then the guidance will
" be by the sea on one side and the upper layer on
the other. The original contributions of Ken-
" " nelly and Heaviside have been republish ed
recently.
By the year 1912, G. K. Pierce and L. De
Forest were discussing in private correspondence the probable explanation of radio signal "fading" in terms of interference between a ground wave
and a sky wave. It is evident that the general
nature of the phenomenon was well understood
at that time, although there mas some dinerence
of opinion in regard to the exact path of the
indirect ray.
In his early publication, Kennelly did not
attempt to discuss the mechanism of atmospheric

"B. Stewart, Encyclopedia Britannica, ninth edition,

p. "1A81.,S1c8h7u8s.ter, Roy. Soc., Phil. Trans. A180, 467 (1889);

Ro'9y.AS. oEc..,KPehninl.elTlyr,anEs.leAc.20W8,or1ld633{91, 940737)(.1902).

~'O. Heaviside, Encyclopedia Britannica, tenth edition,

Vol. 33.

» ~

K. F.

%. Wagner, E. N. E. Smith, J. Inst.

T. 8, 515 (1931).,
Eler. Eng. wireless

section

9,

38 (1934).

conduction, but justihed his assumptions by referring to Thomson's" measurements of the conductivity of air in discharge tubes. Two important advances were therefore made by
Eccles'4 "in 1912.At that time Eccles discussed
the ionizing effect of solar radiation, and also presented the fundamental theory of ionic re-
" fraction. This phase of the theory was extended
by Salpeter" and Van der Pol. The next
" major advance was made in 1924, when
Larmor" reexamined the entire problem and ascribed the major part of the ionic refraction to the presence of free electrons in large numbers.
The possibility of refraction in the Lower atmosphere, as a result of barometric gradient, water-vapor gradient, or temperature inversion, was also examined by several investigators. These e6ects were computed, and it soon became evident that the maximum amount of bending which could be produced by the lower atmosphere mould be insufhcient to account for long distance radio transmission. This phase of the matter has been summarized by Fleming" and by Larmor 2s
" Meanwhile, a simple empirical equation,
known as the "Austin-Cohen formula, had obtained universal recognition as a basis for
engineering design. It resulted from the analysis of a series of experiments on over-water trans-
mission carried on under the supervision of
Austin" in 1909 and 1910. On the cruisers
Birmingham and Salem, numerous quantitative observations of signal strength vs. wave-length mere made in cooperation with a powerful land
station at Brant Rock, Massachusetts, while the ships moved away from the hxed station. The
extreme distances mere of the order of 1200
miles. Most of the original measurements were
made on wave-lengths of 1000 meters and 3750
meters.

~' J. J. Thomson, Recent Researches in Electricity and

Ma~"gWnWeti..smHH..

(Oxford, 1893},p. Eccles, Proc. Roy. Eccles, Electrician

101. Soc. 79,

A87, 79 (1912). 1015 (1912}„Jahrblch

der drahtlosen Telegraphic 8, 253 and 282 (1914).
"J.Salpeter, Physik Zeits. 14, 201 {1913}.
"B.J. Van der Pol, Dissertation (Utrecht, 1920}.
8 Larmor, Jahrblch der drahtlosen Telegraphic

25,

141 (1925).

J. ~' A. Fleming, PrincipLes of Electric Wave Telegraphy

and Telephony (1919),p. 660.

30 L. W. Austin, National Bureau of Standards, Bulletin

7, 315 (1911).

HARRY RO%'E M I M NO

Austin's experimenta results were adequately
described by the empirical relation:
I~ 4—25I. s(h, hg/) d)e

where Ig =current in a 25 ohm receiving antenna of height
h~ kilometers,
I8=current in a transmitting antenna of height
h& kilometers,
d = distance in kilometers, t = wave-length in kilometers,
a=0.0015 for transmission over sea water.

Fleming" shows that the 6rst part of this

equation is entirely consistent with the original

equations of H. Hertz, and represents a simple

inverse-square-law decrease of energy in the

wave front. Austin added the exponential factor

" in order to 6t the experimental curves, and he
ascribed this to "atmospheric absorption. The

quantity X &, which occurs in the exponent, was

introduced by L. Cohen, after a careful analysis

of Austin's original data.

' The theoretical formula deduced from diArac-

tion theorya

contained a somewhat similar

exponential term. However, in addition to a large

discrepancy in the numerical value of a, the

diffraction formula involved the term X & in
place of ) —&. The experimental data seemed

precise enough to preclude this substitution.

Hence the "reQecting layer" hypothesis received

considerable additional support in 1919 when

Katson32 derived the Austin-Cohen formula by

solving a dificult mathematical problem based

on wave propagation in a medium bounded by

concentric conducting spheres. At a later date,

Kenrick" indicated that Watson's result would

not be invalidated if the boundaries of the con-

ducting spheres were not sharply dehned.

For a number of years, all new experiments,

involving additional frequencies, longer dis-

tances, and various types of overland trans-

mission, seemed merely to give greater support

to the Austin-Cohen formula and to extend its

range of usefulness. Immense sums were spent

in constructing powerful long wave stations with

multiple lines of antenna towers ranging up to

800 feet in height All wave-lengths below 200

meters were considered practically worthless for

long distance communication. Consequently this

"J.A. Fleming, Principles of Electric 8"ave Telegraphy
and Telephony (1919},p. 650. 3' G. N. Watson, Proc. Roy. Soc. 95, 546 (1919). 3' G. W. Kenrick, Phys. Rev. 16, 1040 (1928).

entire range of wave-lengths was allotted to the
amateurs. Under such circumstances it is not surprising
that the amateur operators were influenced by conventional ideas at hrst and therefore kept their transmitters very close to the 200-meter wave which represented their legal upper boundary. Amateur activities were severely re-
stricted in many countries, and were entirely forbidden in others, but liberal governmental cooperation was overed in the United States by the Department of Commerce.
The subsequent astonishing progress of amateur radio represents an important and most unusual chapter in scientific history. The American Radio Relay League, a strictly noncommercial organization founded in 1914 by the
late H. P. Maxim, became the nucleus of a
Rourishing international society which has an active membership list drawn from almost every
remote corner of the world. It includes men and
women from all age groups and has representatives in nearly every trade and profession. The League maintains a competent technical staff and its technical journal (QST), and handbook have become practically indispensable in professional engineering laboratories.
This remarkable progress accompanied and
resulted from equally unexpected developments in the study of the short wave region, originally
assigned to amateur use. Under favorable conditions erratic 200-meter communication was obtained at distances far beyond the limit predicted by the Austin-Cohen formula. Systematic tests over longer and longer distances were arranged. The latest improvements in vacuum tube
apparatus were immediately incorporated in amateur circuits, while commercial progress was frequently hampered by patent restrictions and conservative economic policies. Finally, in 1921, the American amateurs sent an expedition, equipped with the latest type of receiving device, to the coast of Scotland, and carried on successful one-way short wave transatlantic tests on a prearranged schedule. The tests were repeated during the following winter and occasional two-
way communication was established. It became
increasingly evident that the orthodox formula was not a dependable guide in this short wave
region.

PHYSICS OF THE IONOSPH ERE

FIG. 1. Long wave daylight transmission.
Stimulated by natural curiosity, and by the overcrowded conditions near 200 meters, some of the more venturesome amateurs began to
investigate still shorter waves. To everyone' s
surprise the transmission became stronger and less erratic. Naturally the downward movement was accelerated, and was limited only by the design of vacuum tube circuits for increasingly higher frequencies.
Below 50 meters another entirely new effect, of outstanding importance, was observed and partially explained. Apparently the signal strength
decreased to zero at points relatively close to the
transmitter (say 50 miles) but exceptionally efficient transmission could be maintained, under favorable conditions, between low power stations thousands of miles apart. The outer boundary
" of the "zone of silence" or "skip region" ap-
peared to be sharply marked. A similar "zone of silence" had been noted
previously in connection with ordinary sound wave transmission of loud noises (produced by artillery fire, or by heavy blasting). The occurrence of such acoustic mirages may be explained
in a satisfactory quantitative manner by considering the refraction produced by temperature inversion in the stratosphere. However, the refraction of electromagnetic waves in this region is insufhcient to account for the observed radio mirages. Numerous "skip zone" observations were carried on in 1924 by an amateur group headed by Reinartz" who correctly ascribed this new phenomenon to the eA'ect of the ionized region. These amateur investigations were im-
" mediately given scientihc verification, and were
extended by Taylor and Hulburt. Rukop'~ has given an interesting survey of the early experiments, and has pointed out that the research
3' A. H. Taylor, Proc. I. R. E. 13, 677 (1925).
"J.L. Reinartz, Q. S. T. 9, 9 (1925).
"A. H. Taylor and E. O. Hulburt, Q. S. T. 9, 12 (1925). "H, Rukop, Zeits. f. Hochfrequenztechnik 28, 41 {1926).

departments of various commercial organizations were equipped with suitable short wave trans-
mitters and receivers at an early date, but had
never suspected the possibility of communication over great distances with such existing apparatus.
%hen the great economic importance of these long-neglected wave-lengths was realized, revised frequency allocations were soon made by international agreement, and the construction of new long wave stations was practically abandoned overnight. In the ensuing struggle between rival countries and rival commercial interests the amateurs quickly lost about 90 percent of their former territory, though a few channels were preserved through the friendly e8orts of the American governmental representatives.
Kith these new experimental discoveries, and
with the theory of electronic refraction overed by Larmor in 1924, the long period of random exploration came to a definite end, and a new field of atmospheric physics began to open up. In previous years physicists had given casual attention to results obtained as a by-product of radio communication. During the succeeding decade most of the advances were made by direct physical measurements, in which the radio apparatus served merely as an incidental research tool. Unfortunately this altered situation has received no recognition whatever in the monumental structure of American governmenta1 regulations. Academic scientific institutions are badly handicapped by inAexible rules which became obsolete twelve years ago. As this situation constitutes a major obstacle, without parallel in other branches of experimental physics, it is necessary to examine the matter in greater detail in Section Q. Supervisory oKcials are aware of the present difficulties and have oA'ered to coc perate, but the cumbersome machinery can be accelerated by a more general knowledge of the facts.
C. BASIC EXPERIMENTAL FACTS
Under this heading I shall attempt to present a
simple analysis of the more prominent features which distinguish the various parts of the radio spectrum, reserving for later treatment the detailed examination of particular experiments

HARRY ROTE M IM No

1'IG. 2. Broadcast wave transmission at night.

designed to test some special aspect of the ionosphere theory. In sketching this general background the picture may be clari6ed by dividing the radio spectrum into defjnite sections which exhibit characteristic types of behavior. A simple quantitative classification of this sort may be useful for illustrative purposes even though it necessarily involves somewhat arbitrary allocations of boundaries betv een diHerent regions of the spectrum. With this reservation concerning the numerical data we may adopt the
following terminology:

I ong waves 'Broadcast waves
"Short waves" A8 C Quasloptlcal vfaves "Microvraves'"

WAvE-LENGTH IN METERS
30,000—600 600-200 200-100 100-50 50-10 10—1
1—0.1

FREQUENCV IN KILOCVCLES
10-500 500-1500 1500-3000 3000-6000 6000-30,000 30,000-300,000 300,000-3,000,000

Ke shall also need to consider several diHerent

concentric regions or "layers" in the ionosphere

which aHect various parts of the radio spectrum

in somewhat different fashions. Adopting Apple-

ton's alphabetical notation, and neglecting all

"6ne structure" for the moment, we may dis-

tinguish three main regions which govern most of

the observed eHects:
— I' region strongly ionized. . . .approximate — E region moderately ionized. approximate — D region weakly ionized. . . .approximate

height, 240 km height, 100 km height, 50 km

After sunset the ionization slowly decreases in
each region, and the weak D region appears to be relatively ineHective at night. The normal
diurnal cycle of ionization and recombination is often modified by sudden and erratic increases in
ionization which may occur in any region at any
hour of day or night. Such changes are especially
common in the D and E layers. The 11-year
sunspot cycle apparently aHects the average ionization in a11 parts of the atmosphere. The

stated heights merely denote representative values chosen to illustrate the order of magnitude. A more detailed discussion of the measured
"heights" will be given in Section K. Free elec-
trons in the Ji region, set in motion by the radio waves, have a comparatively long mean free path, and lose little energy in collisional friction. In the D region, collisional friction is the controlling factor.
"Long waves" obey the Austin-Cohen formula rather well and the concentric-conductingsphere type of transmission offers a reasonable and adequate explanation of their behavior. In the absence of conclusive experimental evidence, we may tentatively assume that the D region serves as the outer conductor in the case of long distance grazing incidence daylight transmission. In comparison with the wavelength it is probable that the lower boundary of the layer is fairly sharply defjned. Consequently the longer waves do not penetrate the ionized layer appreciably and are not absorbed by the high attenuation which would accompany low frequency transmission in an ionized region. The layer acts like a simple metallic reflector, though considerable absorption doubtless results from the slight residual ionization in the troposphere and lower stratosphere. A moderate decrease in attenuation is noted when the transmission path lies on the dark side of the earth. Under such conditions it is natural to suppose
that the upper boundary shifts to the E region.
Fig. 1 indicates continuous wave fronts, extending from the ionized layer to the earth. This distance is comparatively small when measured in wave-lengths, and there is no com-
" plete separation into a "ground wave" and a
"sky wave. Since most of the energy is carried to the receiver by a single ray, the long waves are particularly suitable for "radio compass" applications, being comparative}y free from the distortion commonly associated with plural
Fro. 3. Plural path sky wave transmission.

PHYSI CS OF THE IONOSPHERE

path transmission. The electric vector is approximately vertical, having a slight forward tilt which depends upon the amount of energy absorption in the soil or water traversed. At these low frequencies the ground attenuation is
" relatively small. The effect of ground conduc-
tivity has been discussed recently by Dean. In the case of long wave transmission over sea water Yokoyama and Tanimura" find some evidence which suggests a zigzag ray, reRected successively by the ionosphere and by the sea.
The "long wave" region provides a limited number of high grade transoceanic telegraph channels, free from service interruptions except during severe magnetic storms or abnormally heavy "static" disturbances. The use of "long waves" for telephony is not common, as the
wider frequency bands occupied by telephone channels would cause excessive crowding. Most of the telegraph channels are occupied by expensive high power stations which have been in steady use for many years. In planning new construction the relative stability and reliability of "long wave" transmission does not ordinarily outweigh the decreased cost of a cornparable "short wave" circuit.
"Broadcast waves" provide reliable high quality service in a limited area in the neighborhood of the transmitting station. The dimensions and shape of the area depend upon the antenna design, wave-length, geological conditions, and sunspot cycle, but a circle of 50 mile radius may be mentioned as a representative example. Outside of this area there is a narrow zone characterized by violent fading, which is particularly evident at night. Beyond 150 miles stations of intermediate power are not ordinarily
received during daylight hours, though thousands of miles are frequently covered during the night and transoceanic reception is not unusual. Nocturnal distant transmission is unreliable, however, and subject to moderate fading.
During the daytime nearly all of the energy
" which arrives at the "broadcast wave" receiver
is carried by a "ground wave. The exact nature of this "ground wave" has been the subject of extensive examination. According to the
"E. 'f S. %. Dean, Proc. I. R. E. 17', 1440 {1929).
Yokoyama and I. Tanimura, Proc. I. R. E. 21,
263 (1933).

FIG. 4. "Interaction. "

majority viewpoint the "ground wave" is a true

guided wave, similar in nature to the high fre-

quency waves which can be pre pagated along a

" single copper wire. On the basis of Zenneck's4
diEraction equations, Sommerfeld' developed

a formula describing the transmission of a wave

along the interface between. the atmosphere and

the semiconducting earth. Numerical and graph-

ical evaluations of this formula have been given
" by Hoerschelmann, 4' Ratcliffe and Barnett,
" Rolf, ~ Van der Pol, ~ Kise, 45 Eckersley, " Niessen, '" and Nunier. A separate treatment by

Murray'9 was corrected by Niessen" and shown

to be consistent with the Sommerfeld formula.

Eckersley considers the formula valid for com-

puting direct ray field strengths produced by

waves from 50 meters to 2000 meters at distances

up to 2000 miles.

Bar6eld" has applied the formula in deter-

" mining geological diR'erences, but the short cut
w'hich he uses has been criticised by Englund.

Noether~ admits that the surface waves treated

by Zenneck and Sommerfeld are theoretically

possible, but contends that existing methods do

not produce them. In this connection it is in-

teresting to observe that, in dealing with seismic

waves, Muskat" has recently found it necessary

' A. Sommerfeld, Jahrbuch der drahtlosen Telegraphic 4,

157 (1911);Ann. d. Physik 81, 1135 (1926).

4' Hoerschelmann, Jahrbuck der drahtlosen Telegraphic

5, ~14J.(1A9. 1R2a)t;c5li,ff1e88an(d19M12. )B. arnett, Proc. Camb. Phil. Soc.

23~, 2B88.

(1926). Rolf, Proc.

I.

R.

E. 18,

391

(1930); Ingeniors

Ve~teBns.k.VaAncadde.r

No. 96 (1929). Pol, Tijds. Nederland

Radiogenootschap

"T. 4, 4'~1'0%B5 ..(LV1H9a..n3EW0cdk)ie.esrres,lPePoylr,oacnP.droI.cK.R.I.F.ER. .N.1iE8es,.s21en09,,711A5(n51n59.3{0d1).9. P3h2y)s.ik 5,

273 (1930).

'8 Q'. Nunier, Ann. d. Physik 20, 513 (1934).

"F. '
»

H. Murray,
K. F. Niessen,
R. H. Bar6eld,

Proc. Camb. Phil. Soc. 28, 433 (1932). Ann. d. Physik 15, 810 (1933).
J. Inst. Elec. Eng. 55, 204 (1928).

"C. R. Englund, J. Inst. Elec. Eng. O'7, 931 (1929).

"F. ~

M.

Noether, Muskat,

E. N. T. 10, 160 (1933).
Physics 4, 14 (1933).

HARRY ROWE M I M NO

to introduce a modi6cation in the theory of
propagation of elastic waves along an interface between two homogeneous elastic media.
We could regard the guided wave hypothesis as proven experimentally were it not for the fact that a direct wave, propagated through the troposphere by the shortest route from the transmitting antenna, would be bent into valleys and around obstacles by ordinary diffrac-
tion, and might arrive at nearby receiving points without the help of ground conductivity. This eftect should be computable, but simple knifeedge diffraction is not a suf6ciently good approxi-
mation, and the complete problem apparently has not been solved in a satisfactory manner as
yet. For our present purpose it is sufhcient to describe the "ground wave" as a stable reliable wave which does not show appreciable diurnal variation. Its intensity decreases exponentially with distance. Its range increases with wavelength, and with the power available, and may be increased by designing an antenna which
" concentrates the emitted energy in low angle
radiation.
During the daytime practically all high angle radiation from "broadcasting" antennas is absorbed by the ionosphere. The frequency is high enough to permit the wave to penetrate into the D region where it is rapidly attenuated by collisional friction.
After nightfall the D ionization decreases, and the "broadcast waves" are strongly "reflected"
from the E layer, as indicated in Fig. 2. The
decreased attenuation results from the longer
mean free path of the electrons at the higher altitude. (The refraction and polarization e8ects will be discussed later. ) At points close to the
" transmitter the energy of the sky wave is much
smaller than that of the "ground wave, but the contribution of the "sky wave" does not decrease rapidly with distance. As a consequence of the exponential decrease in the "ground wave" we soon arrive at a zone in which the two waves produce approximately equal 6elds. The slightest change in atmospheric conditions will cause violent fading as the phase relation of the two waves varies. The commonly observed periodic
"H. E. Gihring and G. H. Brown, Proc. I. R. E. 23,
311 (1935).

" fading is an indication of a progressive change in
the equivalent path of the "sky wave. At distant points the "ground wave" is en-
tirely ineffective, but the sky wave does appear
at great distances under favorable conditions. Namba and Hiraga" have observed an im-
provement in propagation across the Pacihc during the years near the sun spot minimum. Such transmission is better in autumn than in midwinter and is worst in midsummer. Berkner'~ has reported on the reception of numerous American broadcasting stations in the south polar district, at distances exceeding 12,000 km. Beyond the ground wave zone the signal is often comparatively steady, though not dependable. Slow alterations in 6eld strength occur as a result of changes in absorption. On other occasions periodic fading results from the type of plural path sky wave transmission indicated
in Fig. 3. The ionized region is not a completely linear
transmitting medium. As indicated in Fig. 4, transmission from point A to point C may be noticeably distorted by cross modulation from a powerful interfering station located near the halfway point on the great circle which connects the transmitter and receiver. These "interaction"
e8'ects will be discussed in Section 5. The dis-
turbance is unimportant except at the low
frequency end of the broadcast band. For convenience in describing the "short
wave" region, j: have divided it into three parts. The wave-lengths from 100 to 200 meters are notoriously erratic and unsatisfactory for distant communication. The ground wave is attenuated so rapidly that the local service area is too small for ef6cient broadcast utilization. The sky wave is also subject to excessive attenuation, and this
" part of the spectrum may be considered as an
actual atmospheric "absorption band. Though other theories will be mentioned later, it is almost certain that these anomalous atmospheric effects are due to the resonance frequency of the free electrons set in motion by the wave. The electrons travel in circular orbits about the lines of force of the earth's magnetic held and have a
natural frequency in this general range.
"S. Namba and D. Hiraga, Radio Research Japan,
Report 2, 9 (1932).
~7 L. V. Berkner, Proc. I. R. E. 20, 1324 (1932).

PHYSICS OF THE IONOSPHERE

FIG. 5. "Skip distance. "
The erratic properties of the 100—200 meter waves arise in part from an additional complication. This is a transition range of frequencies,
and reHections may be expected from the E layer or the F layer or simultaneously from both
layers. The shorter waves in this band are often
able to penetrate the E region, but their fre-
quencies are not high enough to prevent appreci-
able partial reHection, attenuation, polarization
and reduction in group velocity.
In "short wave" group 8, we may conveniently
include the range extending from 50 to 100
meters. These waves are especially valuable for overland transmission within the confines of a
single continent and they are widely used by aviation interests and by the military forces.
Except at points within 30 miles of the transmitter, where the ground wave produces an
appreciable eRect, reception depends upon sky wave propagation. In general the dependable sky wave comes from the I' layer, though
strong E layer "reHections" are not infrequent.
Even in the extreme case of signals received at
nearby points after reHection" from the ionized
region at nearly normal incidence, E layer
"reHections" often occur. These transient "re-
Hections" frequently appear to be caused by relatively small dense ionic clouds, drifting at
random in the E region. On other occasions
there is a general increase in ionization which
shifts the path from the I" to the E region for a
number of hours. As the shift often occurs very quickly, transitions of this sort frequently occur
without severe fading or interruption of service.
8 Though group is sometimes used in trans-
oceanic service the grazing incidence absorption
is somewhat greater than in group C. Group C, extending from 10 to 50 meters, is
chieHy characterized by remarkable efficiency in

long distance intercontinental transmission, and
by the prominence of the skip distance effect, previously mentioned in the historical survey.
In order to understand this "skip distance"
phenomenon let us first fix our attention on the nocturnal Ii layer "sky wave" or echo, which
returns to earth at a point near the transmitter after "reHection" at nearly normal incidence.
As the ionization density slowly decreases during the night the "sky wave" gradually penetrates to greater and greater heights in the ionized
region. The corresponding time lag of the echo signal is gradually increased. A similar effect may be produced artificially by increasing the transmitter frequency while the ionization density remains substantially constant. In either case a critical condition may be attained at which the slow increase of "effective height" (measured as time lag of echo) is broken by a sudden upward surge. At the same time the strength of the "reHected" signal decreases rapidly, and it
suddenly vanishes completely. The sudden increase of time Iag is readily
accounted for as evidence of a corresponding decrease of group velocity in the wave, as it penetrates into and passes through a considerable thickness of ionized I' layer. Similar effects are
frequently produced by the E region, when the
ionization density is near a critical value which will just suffice to produce "reHection" at the given frequency. Such phenomena are predicted by the mathematical theory, and it is natural to expect an associated decrease in signal strength due to attenuation.
By analogy with the behavior observed under
similar circumstances in the E region, it is
possible to assume that the final complete disappearance of the signal corresponds to com-
I plete penetration of the layer and resultant
loss of the signal in interstellar space. This is the customary assumption and it has been regarded as a self-evident fact by many writers. However, Eckersley favors the alternative idea of complete absorption within the I' layer. The
E. matter will be referred to again in Section
According to either hypothesis a signal which has been lost in this manner can be restored by increasing the angle of incidence beyond a definite critical value. Consequently we should expect the ray pattern represented in Fig. 5.

HARRV RGWE M I M NO

At points within a short distance (say 15 miles) of the transmitter a "ground wave" signal appears, but this is completely ineffective at distant points on account of excessive attenuation. At points beyond this small "ground wave"
xone no dependable signal is observed until we
reach a sharply marked boundary curve where the usual downcoming sky wave appears. The
main reception zone lies beyond this boundary. By careful examination these "skip-distance"
phenomena may be detected at night in the
50—100 meter region, but in general such skip
distances are short and the effect is largely
masked by ground wave propagation. In the 10—50 meter band, however, the skip distances
ma, y be measured in thousands of miles. Here the effect becomes the controlling factor in determining the best frequency for use on a
dehnite long distance circuit at a particular
hour, season, and year. The "silent zone" is not completely dead.
Strong signals produce a peculiar reverberant echo, easily recognized by an experienced ear in voice or code transmission. These "scattered" signals are not yet completely explained, but the matter will be referred to in greater detail
in Section R.
As the frequency of the wave is increased the skip zone boundary moves outward until the "silent zone" 6nally embraces the entire earth,
and transmission by means of ionosphere "re-
Hection" is no longer possible. Under average
conditions this limita, tion is reached in the
vicinity of 10 meters. Consequently, wavelengths somewhat longer than 10 meters are of
great value in daylight transmission over very long paths, while wave-lengths slightly shorter than the critical value are quite useless for this
purpose. However, the ionization density of the I' layer varies considerably from day to day. Departures from the average value not infrequently permit freak transmission on wave-
1eagths as short as 8.5 Ineters and it is quite certain that 5 meter signals can be received at
great distances when the ionization is excep-
tionally high.
In general, however, absolutely no dependence can be placed on ionosphere transmission or guided ground wave transmission in the "quasioptical" range, extending from I to 10 meters.

I')G. 6. Relation between actoal path and "equiva)ent" path of ray.
Though apparently narrow when measured in terms of wave-length, this classi6cation evidently covers an enormous range of frequencies, and is therefore particularly attractive for future television applications. These waves are also especially applicable to two-way communication
with police patrol cars, since a short vertica1 rod, easily carried on a moving car, serves as a decidedly efticient quarter-wave transmitting antenna. The dependable service area is practically limited to the territory included within the optical horizon, and these waves are chieRy useful for communication within a single metropolitan district.
Unfortunately, though, the interference area of quasioptical transmission is much greater in extent than the dependable service area. Even with comparatively low power used in present experimental installations, very strong 5 meter signals are often received far below the optical horizon at points more than 100 miles from the transmitter. This effect is dehnitely associated with weather conditions in the lower troposphere and it appears to be a simple mirage phenomenon caused by the temperature inversions which frequently exist in the atmosphere a few thousand feet above the ground level. In fact it appears to be possible to use this type of transmission as a new meteorological tool in studying the distribution and movement of air masses. Preliminary studies, conducted by the American Radio Relay League in cooperation with Harvard University, have indicated a remarkably close correlation between quasioptical transmission over the Boston-Hartford path and the corresponding meteorological temperature gradients ("lapse rates") measured at the Boston Airport
and at Mitchell Field, near New York City, By means of simultaneous held experiments on

I PHYSI CS OF HE IONOSPHERE

wave-lengths of 3.25, 2.50 and 5.00 meters we
are now investigating the possible eHect of hilltop diffraction on quasioptical transmission. On account of excessive attenuation the simple guided wave, following the surface of the ground,
appears to be quite ine6ective at such high
frequencies. %. W. Mumford considers that the
large dipole moment of water vapor molecules may contribute greatly to the bending of quasioptical waves. Water vapor gradients
" usually accompany temperature inversions. Presumably the "microwaves, extending from
O.i to 1.0 meter, will exhibit characteristics
resembling the "quasioptical" waves. Research in this region requires the construction of special types of transmitting tubes. The power outputs and efficiencies, thus far obtained, are very low, and little information is yet available on transmission. However, a successful two-way 17 em channel has been established across the English Channel and other experiments are in progress.
D. ELEMENTARY THEORY
The detailed study of the action of the ionosphere requires an intricate mathematical formulation, but the basic facts may be explained in a very simple way.
Consider the action of a single electron placed between the plates of a condenser. If the frequency of the alternating voltage applied to the condenser plates is low, and if the motion of the electron is restrained by elastic forces, the displacement of the electron is substantially in phase with the applied voltage, and the vibrating charge constitutes an alternating current, in phase with the Maxwell displacement current through the otherwise empty space. In other words, the negative electron approaches the positive plate of the condenser at the peak of
the positive half-cycle. It therefore neutralizes
a portion of the charge on the condenser plates and permits the available voltage to drive a larger charging current through the external circuit. Under such circumstances we are accustomed to say that the presence of the bound electron has increased the dielectric constant of the region permeated by the electric 6eld.
On the other hand, if the elastic restraining force acting on the electron is zero, or if the

frequency is so high that the inertial force
predominates, the phase relations reverse, a,nd
the vibrating charge may reduce the eR'ective dielectric constant of the medium below unity. We are familiar with this effect in optics, since it provides the conventional elementary "resonance" explanation of the anomalous dispersion
observed on the high frequency side of an
absorption line in the optical spectrum. If present, collisional friction introduces a re-
sistance term which decreases the magnitude of the change in dielectric constant but cannot alter the sign.
Precisely the same facts determine the basic conditions governing the propagation of a radio wave in the ionosphere. Due to the action of free electrons, set in motion by the electromagnetic field of the wave, the dielectric constant
and the corresponding index of refraction of the ionized region are less than unity. Consequently the layer is a medium which is "less dense" in
the optical sense than the nonionized strata of air beneath. If the boundary between the media
is reasonably sharp (in comparison with the wave-length of the electromagnetic radiation)
total internal re Rection takes place at the boundary, and rays which strike the layer at
angles exceeding the critical angle will be strongly reRected back to the earth.
If the boundary between the two media is not,
sharp the incident ray is gradually refracted farther and farther away from the normal, describing a curve which depends upon the rate of increase of the free-electron density. If the total increase of electron density sufFices, the ray wi11 eventually attain a horizontal direction and will then follow a symmetrical downward
path which brings it back to the earth. If the
maximum density is insufhcient the ray will penetrate the layer. For a layer of given maximum density (and a wave of given frequency), there is evidently a definite critical angle of original incidence, which determines whether or
not a given ray will return to earth. The study of this type of refraction may be regarded as a
detailed examination of the mechanism which is responsible for the "reAection" of an electromagnetic wave from a conducting surface. When the boundary of the conducting "layer" is not sharply de6ned, it is sometimes advantageous

HARRY ROWE M I M NO

to consider the actual curved path of the ray. For many purposes, however, it is sufficient to
" replace the curved path ABCDE by the angular
"equivalent path, ABC'DE, represented in Fig. 6, and to speak of the "equivalent height" of a fictitious "reAector. "' We may use the term "actual height" to refer to the height at which maximum free-electron density occurs. In the
" special case of a ray which strikes the layer at
the "critical angle, this height coincides with the maximum height of the curved ray. All other "totally refiected" rays will fall beneath
this height.
Breit and Tuve" have called attention to the fact that the time required for a definite radio signal to traverse the route ABCDE, through the actua1 ionized medium, is exactly equal to
the time which would be required for the signal
" to follow the route AJ3C'DE in empty space.
Consequently the "equivalent height, defined
geometrically by Fig. 6, is properly measured by direct observations on the time lag of echoes, without correction for the reduction in group velocity which delays the signal in the ionized layer. Equivalent heights measured in this fashion are therefore consistent with equivalent
heights determined geometrically by a measurement of the angle of arrival of the downcoming
I ay. Consequently the "actual height" of a layer
and its ionization gradient cannot be measured directly by a single observation of any sort. These quantities must be inferred indirectly from a series of observations of "equivalent height" using diA'erent frequencies or different paths or both. There is reason to believe that the
lower boundary of the E layer is relatively
sharply defined and that the measured "equivalent height" exceeds the "actual height" by
only a few percent. The F region is more disuse
and the "actual heights'" are not well known as yet. Preliminary data suggest that the "equivalent heights" may exceed the "actual heights" by at least 25 percent. When not more explicitly defined, statements in the literature in regard to layer "height" nearly always refer to the "equivalent height" determined by direct experimental observation.

Disregarding, for the moment, modifications produced by the earth's magnetic field and by collisional friction, we may readily deduce a simple formula for the index of refraction of an ionized medium. A more elaborate analysis will
be given in Section H. Pedersen" has suggested
the following simplified treatment, based on
" elementary considerations, which yields the
familiar equation of Eccles" and Larmor. Consider a condenser with plates of unit area,
separated in vacuum by a distance of 1 cm. When provided with suitable guard rings, the
unit condenser has a capacity of 1/4' absolute
units. Now introduce N electrons of charge e and
mass m in the cubic centimeter included between condenser plates and apply an alternating voltage of instantaneous value v, of angular frequency co,
I and amplitude V. The electrons will be set in
motion by the electric field, and their velocity will lag a quarter period behind the applied voltage and will have a maximum value
U= (e/a)m) V.

But N charges of magnitude e, moving with
the common velocity I, are equivalent to a

current element of 1 cm length and with the

strength

i, =Re u.

The amplitude of this lagging alternating current is therefore

I,.= Ne U= Ee V=

and the vibrating electrons produce the same
effect as an equivalent inductance of m/¹'
absolute units, shunted across the condenser. But so far as the total current in the external
circuit is concerned this combination is equivalent to a condenser with the capacity
C' = 1/4~ —(¹'/cv'm)
and therefore to the unit condenser with the dielectric constant
e = 1 —4s ¹'/a)'m.

"G. "P. Breit and M. A. Tuve, Phys. Rev. 28, 554 (1926).

O. Pedersen, Wireless Engineer V, 16 (1930).

PH YSI CS OF TH E IONOSPHERE

In this equation eu and N may be regarded as independent variables, though only co is under experimental control.
It is evident that the theory also predicts an
additional reduction of dielectric constant due to the presence of heavier ions of either sign, and we could readily introduce additional terms of similar algebraic form to take account of this effect. However, as the mass of the ion occurs in the denominator of the subtractive term, a single electron is more eHective than a vast number of heavy ions. Nevertheless, some physicists believe that hydrogen ions may produce a detectable effect in the ionosphere. This possibility will be referred to in a later section.
It is also evident that the increase in dielectric
constant which occurs during the night may result mainly from neutralization of electrons by positive ions, or mainly from mere attachment of electrons to heavier neutral atoms or molecules.
Since N may become very large it is evidently possible for the dielectric constant to be reduced to zero and even to be reversed in sign. Such
action commonly occurs in the ionosphere in the frequency ranges used in practical communica-
tion. It is therefore important to examine the
e6'ect on the index of refraction of this reversal of the sign of the dielectric constant.
Pedersen points out that the usual approxi-
mate expression, n = Qe, is valid only for positive
values of ~. This is an important restriction as a number of writers have assumed that a negative dielectric constant necessarily implies an imaginary refractive index. In a nonmagnetic conducting medium the correct relation is
n = Le/2+ I e'/4+ (2mc'o/(a)'I &g&.
Where the conductivity may be neglected this reduces to

which gives

n = Qe m =0

for for

e~0,
e —0.

We may also note that the phase velocity,
U„,of the wave in the ionosphere is given by the
expression
V„=c/(1 4s.Ne'/co'—m )&,

where c is the velocity of light.

The velocity V~ (which exceeds the velocity of light) is not a directly measurable quantity. The speed of travel of the observed radio signal is measured by the group velocity, U„where
( ru dV„q V„d(a)
= c(1 —4m ¹'/oem) &.
The signal is therefore retarded by the presence of free ions in the medium.
When applied to the ionosphere these considerations have direct application to the most prominent experimental facts encountered in ordinary communication, though we must later consider the modifications caused by the magnetic field of the earth and by collisional friction.
As the index of refraction of an ionized layer approaches zero the critical angle also approaches zero. In the limit we might expect that total
internal reflection would occur even at normal
incidence. Ke may express this prediction in
other language by saying that the "skip zone" should shrink to zero for a suSciently high electron density in the layer or a su%ciently low frequency at the transmitter. These expectations are amply verified by experiment.
It is therefore customary to say that a radio
signal, directed vertically upward, penetrates into a region of increasing free-electron density and travels with a slower and slower group velocity, until it finally reaches an altitude where the electron density is just sufhcient to reduce the index of refraction to zero. At this point the direction of the ray is reversed by total internal refiection and the signal is propagated
downward at an increasing speed, attaining the velocity of light as it emerges from the ionized layer. If the concentration of free electrons at
the most dense portion of the layer is not quite sufFicient to reduce the index of refraction to zero, the vertical ray passes completely through the region and the signal emerges above the layer with the velocity of light. At some greater height it may then encounter a denser layer which will return it toward the earth, and it passes through the lower region a second time on its way to the receiving antenna. When the maximum electron density in the lower layer

HAR R Y ROWE M I M NO

is only slightly less than the critical value, the signal is considerably retarded by this round trip through the lower stratum and the measured "equivalent height" of the upper layer may be
abnormally increased by at least 100 percent.
Such effects are readily recognized and interpreted, however, when they occur in a series of experimental observations.
Though this elementary explanation of vertical incidence "reHection" in a di8'use medium is strongly suggestive and does agree well with the experimental observation, it cannot be regarded as a rigorous and complete theory. We have no apparent right to generalize the simple laws of total internal reHection and apply them so confidently in a region which has continuously variable optical properties. Various attempts to resolve this difhculty will be referred to in Section G, but the complete theoretical problem is a difficult one which apparently has not yet been solved in a manner which is thoroughly satisfactory.
E. THE NATU'RE OF THE ACCELERATING FIELD
Though the elementary theory, presented in Section D does correlate a number of interesting experimental facts, we shall need to invoke a more elaborate mathematical treatment in order to explain the double-refraction eAects which are frequently observed. As a preliminary step in considering the more advanced theory we shall 6rst examine several fundamental postulates in regard to the forces which determine the motion of the electrons.
Fabry" has called attention to the fact that
free electrons are quite eR'ective in scattering radio waves, since phase addition is to be ex-
pected at these comparatively low frequencies. In the case of free electrons set in motion by a
light wave, only the intensities add. Ke should
therefore expect little resultant modification of the light beam, though it is possible that slight astrophysical effects may be detected.
However, when we attempt to compute scattering of radio waves resulting from electron motion we immediately- become involved in a serious controversy in regard to the nature of the
~o C. Fabry, Comptes rendus 187, 777 (1928),

accelerating field. In the ionosphere, in addition to the free electrons, positive ions and neutral molecules are present in large numbers. In the elementary treatment presented above, and in the theoretical papers which first appeared, the simple space-average value of the electric 6eld was employed, without critical examination, in computing the motion of the individual electrons. When investigating the alteration of dielectric consant due to the presence of ordinary neutral atoms between condenser plates, we know that this "pipe force" must be corrected by the Lorentz "polarization" term which takes account of the actual corpuscular distribution of matter. We often visualize the individual atom or molecule as an object enclosed in an approximately spherical cavity, influenced by the field due to induced charges on the walls. This physical picture may be used in order to compute the Lorentz "polarization" term, though several other methods of computation lead to the same mathematical expression. The resulting mathematical formula has been tested by experiments upon the dielectric constant of ordinary materials, and good quanti-
" tative agreement has been obtained. Consequently, Hartree" included this Lorentz correction term in several mathematical
papers on the scattering of radio waves in stratified media, and emphasized the fact that Goldstein" and previous writers had omitted
it. The corrected expression has been employed
" by Appleton" '~ in several papers, and the
resulting "Appleton-Hartree formula, for the dielectric constants of a doubly-refracting medium; has been applied extensively in numerical
calculations by Taylor" and numerous others. The numerical correction is a large one, which alters by 6fty percent the value of the electron
density computed from measurements of penetration frequency. The actual equations will be given later.

"D. ~
~ ~

R. Hartree, Proc. Camb. Phil. Soc. 25, 97 (1929). D. R. Hartree, Proc. Camb. Phil. Soc. 2V, 143 (1931) D. R. Hartree, Proc. Roy. Soc. A131, 428 (1931). S. Goldstein, Proc. Roy. Soc. A121, 260 (1928).

"E.V. Appleton, J. Inst. Elec. Eng. Vl, 642 (1932).

"E. V. Appleton and R. Naismith, Proc. Roy. Soc.

131, 36 (1932).

~~ E. V. Appleton and G. B. Builder, Proc. Phys. Soc.

45, 208 (1933).

6 Mary Taylor, Proc. Phys. Soc. 45, 408 (1934).

PHYSICS OF THE IONOSPHERE

" In f933, Tonks" questioned the validity many of those always accepted in theoretical

of Hartree's derivation and contended that the physics" it is possible to support either of two

polarization term should be omitted except entirely contradictory formulae. Though the re-

" where "there is some detailed arrangement of the examination of these older viewpoints does not
negative with respect to the positive charges. provide a clear-cut answer to the question, the

In this objection he was supported by Norton" general weight of evidence favors the omission

who maintained that the space-average 6eM is of the Lorentz term when computing the force

applicable, since the electron moves over a which acts on a free electron in the presence of

distance which is large, with respect to the elec- positive ions or neutra1 atoms and molecules.

tron spacing, in a time which is short, in com- Since atoms cannot interpenetrate, an individual

parison with the period of the wave.

atom is a6ected only by the external 6elds of

Though the papers by Tonks and Norton other atoms. This restriction does not apply to

started a lively theoretical argument there was the free electron and the simple space average

no immediate agreement in regard to the merit should be used in determining the force.

of the various viewpoints which had been pre-

However, Darwin obtains a more decisive

sented. Hartree~ pointed out two unjusti6ed formulation of the problem by an entirely new

assumptions in the arguments offered by Tonks, method which avoids the troublesome am-

and one such assumption in his own previous biguities involved in the close ana1ysis of internal

derivations. The important question at stake electric fields in matter, and he supplies condi-

remained undecided. In self-defence several tions for discriminating between substances

experimental physicists formed the habit of requiring the two types of formula. By the

publishing two sets of numerical computations, application of Hamiltonian dynamics it is

allowing the reader to take his own choice.

possible to formulate the electric moment of, a

In 1934 Darwin published a preliminary small volume of the medium, the size of the

paper, ~' indicating that the Lorentz term should region being such that the retardation of the

not be included. In December, 1934, he presented waves is negligible. The conditions for the

an extensive and thorough reexamination'4 of omission of the polarization term are fulfilled in

the entire problem which appears to settle the the ionosphere and in metals. VVhen applied to

matter decisively. In his recent paper Darwin the determination of the properties of metals,

notes that the problem is a perfectly definite one and it should be possible to solve it completely without resort to experiment. The main

the theory is supported by quantitative experimental evidence, and it appears to aRord a reliable basis for the quantitative computation

work on the subject was done more than fifty of the properties of the ionosphere.

years ago, A purely classical treatment is

suf6cient, though quantum mechanics methods P. OTHER FORCES ACTING ON THE ELECTRON
may be employed without altering the result.

Though superficially simple, the problem is Having discussed the nature of the electric exceedingly treacherous. Since the average 6eld 6eM produced by the incident wave, we may

is very small in comparison w'ith the 6elds which next consider several additional factors which

exist at local points in the medium the computed may affect the motion of the scattering electrons.

value may be greatly aHected by a slight alteration in the method used for evaluating the
average. Darwin 6rst considers the various

The frictional force due to collisions and the
deAecting force produced by the earth's magnetic
heM will be treated mathematically in Section G.

methods which have been employed previously. Perturbing forces caused by interfering signals
By means of arguments "quite as convincing as will be described in Section S. In addition to

"L.Tonks, Nature 132, 10j. (1933).

"L.Tonks, Nature 1M, 710 (1933).

"K. A. Norton, Nature 1M, 67'6 (1933).

"»

D. C.

R. G.

Hartree,
Da~in,

Nature Na~~r~

1N, 929 (1933). 133, 62 (&934).

"C. G. Darwin, Proc. Roy. Soc. 145, 1'7 (1934}.

these effects we must examine a hypothetical "quasielastic" or "relaxation" force which has come into the literature as a result of laboratory experiments designed to test the properties of
ronrzed gases.

HAH. R. V B. OWE M I M NO

" In 1913, Salpeter" developed equations for quencies above resonance the dielectric constant

the refraction of electric maves in an ionized would be reduced and refraction could occur.

gas. Though similar to the treatment given by For lower frequencies transmission could be

Eccles24 in the previous year, Salpeter's deriva- ascribed to conductivity only.

tion included a more detailed consideration of Though this ionosphere extension of the theory

collisional friction. In 1920, Van der Pol'~ per- was never widely accepted by other investi-

/'ormed an experiment on an ion plasma between gators it was generally admitted that Gutton

the plates of a condenser and attempted to and Clement had undoubtedly made accurate

verify the theory of Eccles and Salpeter by observations of the alterations in the apparent

direct test. He sought to measure a change in capacity of their laboratory circuit, and many

condenser capacity resulting from the formation persons believed that "quasielastic" forces

of ions, and for this purpose he employed the might account for some of the ultra-high-

Lecher wire technique suggested in 1897 by Drude. 7' For certain adjustments of the apparatus, Van der Pol obtained a qualitative indica-

frequency oscillations which occur in gaseous
" electron tubes. H. Gutton" believed that such
forces produced the "plasmoidal oscillations,

tion of a reduction in the dielectric constant.

treated theoretically and experimentally by

In 1927, H. Gutton and G. Cement~' ~9 Tonks and Langmuir" though he concluded that

repeated this experiment with slightly different the theory was not precise enough to provide

experimenta1 technique and tried to obtain im- quantitative verification.

proved quantitative data. As predicted by the In 1927 and 1928, Pedersen' and Rybner"

theory, they did obtain a decrease in the di- claimed that the apparent anomaly was entirely

electric constant when ions were formed. How- due to the apparatus which Gutton and Clement

ever, when they increased the ionization beyond had used. In j.929, Bergmann and Diiring"

a certain critical value, the dielectric constant repeated Van der Pol's original Lecher wire

apparently suddenly increased, in contradiction experiment but substituted electrons liberated

to the expected behavior. This critical value of from a hot cathode in place of the ionized gas

the ion density could be reduced by performing used by previous investigators. Consequently

the experiment with a resonant circuit of lower the mean free path of the electrons was large.

natural frequency.

Under these conditions they obtained complete

Gutton and Clement attempted to explain this quantitative agreement with the simple theory

anomalous behavior by assuming that the within the limits of accuracy of the measure-

vibrating ions were subjected to a "quasi- ments. Using a magnetron tube, Benner" like-

elastic" restoring force produced by the mutual wise verified the simple magneto-ionic theory

action of the ions. The ion plasma would there- regarding resonant changes in dielectric constant

fore have a natural period of oscillation which and decrement of an ionized region.

mould increase with the density of the ionized

In a separate paper Benner" offered an im-

gas. By extrapolating similar laboratory measure- proved formula for the Bergmann-DOring ex-

ments C. Gutton'0 sought to show that such periment, which better suited the special experi-

gaseous resonance would occur in the ionosphere mental conditions involved. In his original

at a frequency in the neighborhood of 1500 kc derivation Salpeter had assumed oscillations

and he considered this to be the explanation of interrupted by collisions, whereas Bergmann

the poor transmission obtained with wave- and During had made use of a mean free path

lengths of the order of 200 meters. For fre- much larger than the amplitude of oscillation.

J. ~' Salpeter, Jahrbuch der drahtlosen Telegraphic 8, 247
"P. (1914). Drude, Ann. d. Physik 61, 466 (1897). ~' H. Gutton and J. Clement, Comptes rendus 184, 441
(19"2H7).. Gutton and J. Clement, Comptes rendus 184, 676 "C. {1"9H27.).Gutton and J. Clement, Onde Elec. 5, 137 (1927).
Gutton, Ann. de physique 14, 5 (1930).

8~
~

H.
L.

TGountktosn,andAnIn..

de physique 13, Langmuir, Phys.

98 {1930). Rev. 33, 195

(1929).

3 P. O. Pedersen, The Propagation of Radio IVaves

(Co~'pJLen..hRaBygbeernng,emr,a1nO9n2nd7e)a.ndfleeW. .'7,

428 (1928).
During, Ann.

d. Physik

1,

1041 {1929).

8' S. Benner, Naturwiss. 1'7, 120 (1929).

8'S. Benner, Ann. d. Physik 3, 993 (1929).

PHYSICS OF THE IONOSPHERE

In a general survey article on the interna1 action to a large positive value. Though this critical

of thermionic systems„published in 1931, condition is purely a property of the electric

Benhams' criticized this revised formula, but circuit itself, the eHect produces a striking

Benner" promptly pointed out that the equation imitation of a resonant change in the properties

had been misused by Benham.

of the ionized gas.

In 1932, Niessen' published a theoretical

Though Appleton considers the Gutton "quasi-

paper indicating that, in the ionosphere, the resonance" entirely spurious, he does verify

attenuating eA'ect of a "relaxation force" could certain experimental observations made by

be neglected in comparison with collisional Tonks~ which indicate the presence of a second

friction. "Relaxation resonance" would be im- "resonance" point, obtained with low values of

possible. If present, however, the "relaxation tube current. Appleton considers that Tonks and

force" would always cause absorption and might also exert an apparent binding action on the free electrons.

Langmuir are not justified in describing this
" e8ect as "plasma-resonance, however. He pre" fers to regard it as "sheath-resonance, which

In 1932, after completing experiments begun takes place at the boundary of the plasma,
by Appleton and Childs, 9' Appleton and Chap- since "it can be shown that an electron vibrating

man" announced a thorough experimental re- from the main discharge into a sheath will be

" examination of the Gutton-Clement investiga- subjected to a restoring force and thus have a
tion. A supplementary report of this test was natural vibration frequency. Though important

presented by Appleton" at the 1934 Congress in the theory of discharge tubes, sheath-reso-
of the U, R. S. I. After duplicating the apparatus nance may evidently be disregarded in the

used by Gutton and Clement, Appleton and ionosphere.

Chapman checked the original experimental

In France, the concept of "quasielastic" forces

" observations but denied all of the conclusions. still receives strong support at the Laboratoire
There were no '"quasielastic" forces. All "quasi- Nationale de Radiohlectricith. C. Gutton, di-

resonance" effects were due to the fact that the rector of the laboratory, recently presented an

dielectric constant assumes negative values. The interesting paper summarizing the French experi-

explanations previously overed by Pedersen and ments, but apparently he has not refuted Apple-

Rybner were verified quantitatively. The con- ton's quantitative data as yet.

denser containing the ionized gas is not a simple

If we therefore reject the idea of "quasi-

structure with a uniform dielectric. In part of resonance" in a plasma, we may still conduct a

the space permeated by the electric field the search for real resonance frequencies due to the

dielectric constant is positive. The stray capacities of the coil and wiring must also be con-

internal Ziegler

'

vibrations summarized

of

large molecules. the theories and

In 1934, measure-

sidered. %hen describing the actual distributed ments of dielectric constant and concluded that

circuit by an equivalent simple circuit with there is no satisfactory evidence of any such

lumped constants, it is useful to consider Imo effect in the ultra-high-frequency radio spectrum

condensers in series with the coil. In one of these condensers the dielectric constant can assume negative values, but the other condenser contains a normal positive dielectric which is not aRected by the ionization. As the ionization density increases the resulAxnt series capacity assumes a large negative value and then suddenly changes

extending from 100 cm down to 10 cm. At a
wave-length of 1.1 cm however, Cleeton and
VA'11iams'~ have demonstrated an absorption
effect due to the "turning-inside-out frequency" of the ammonia molecule. This type of research will undoubtedly be extended but molecular resonance may be disregarded so far as iono-

88 W E. Benham, Phil. Mag. 11, 451' (1931). '9 S. Benner, Phil. Mag. 11, 1252 (1931).
"K. F. Niessen, Physik Zeits. 33, 705 (1932).
"E. 969 {193V0.).,Appleton and E. C. Childs, Phil. Mag. 10, "E.V. Appleton and F. %'. Chapman, Proc. Phys. Soc.
44, 246 (i932).
'~ E. V. Appleton, U. R. S. I., London Congress {1934).

sphere refraction is concerned.
""L.Tonks, Phys. Rev. 37, 1458 (1931). C. Gutton, U. R. S. I., London Congress (1934).
"C. O' W. Ziegler, Physik Zeits. 35, 476 (1934).
E. Cleeton and N. H. %illiams, Phys. Rev. 45,
234 (1934).

HARRY ROKE M I M NO

G. THE EqUxvrows OF Monow OF vHE
EI.En Rows

Theoretical investigations, veri6ed by direct experimental tests, indicate that the magnetic

6eld of the earth is responsible for the doublerefraction eRects frequently encountered in commercial radio transmission and in ionosphere research. The experiments are precise enough to verify the known quantitative facts in regard to the variation of magnetic 6eld intensity with latitude, and to provide means for determining the variation with height within the earth' s

atmosphere.

In the special cases of transmission along the magnetic 6eld and transmission perpendicular to the magnetic 6eld this eR'ect was studied in
I926 by Appleton and Barnett" and inde-
pendently by Nichols and Schelleng. Below 70—80 km electrons collide so frequently with gas molecules that the magnetic 6eld is relatively ineRective. Above this critical height the collision
frequency is lower than the rotational frequency about the earth's held. Consequently the in-
ductivity in the direction of the 6eld is appreci-
ably different from the inductivity at right angles to the earth's 6eld, and the plane of polarization is rotated. The Faraday eRect and the Kerr eRect are produced according to the
wave-length. So long as the investigation was limited to the two special cases mentioned above it was possible to make use of the equations previously employed by Voigt coo Drude co and Lorentz'" in the study of the propagation of optical waves in the held of a magnet.
Appleton'" and a number of others soon ex-

tended the equations of magneto-ionic double refraction in order to include the general case of propagation at an arbitrary angle with respect
to the direction of IS. The appearance of the
resulting expressions for the refractive indices

depends greatly on the selection of the directions of coordinate axes and the nature of the abbreviations used in reducing the cumbersome ex-

98 K. U. Appleton and M. A. F. Barnett, Nature 115,

333"H(1.92%S'.}N;Pircohco.lsRaonyd.

Soc. 109, 621 (1925).
J. C. Schelleng, Bell Sys.

Tech.

J.

'" 4, 215 (192S). W. Voigt, Mggneto end Elektro-optik (Leipsig, 1908).

1I
'~ '~

P. H. E.

(192'l).

E}rude L8$t'gsgk get' optfk (Leipslg, 1QOO). A. Lorentz, Theory of Eledmws (Leipsig, 1909).
V. Appleton, U. R. S. I., Washington Assembly

pressions to a compact and useful form. Although

the analysis is already available in the scienti6c

literature of several different countries, the im-

portance of the eRect justi6es the repetition of

the derivation in outline form at this point.

As Appleton's'~ adaptation of the Lorentz

equations has been widely used in numerical

computation, I shall adopt his notation and

choice of axes, but shall eliminate the Lorentz

correction term for the reasons given in Section
E. The reader may also wish to study the deriva-

tions, presented Forsterling and

in somewhat different Lassen 'o~ Baker and

form, Green

b' y'

and Pierce. "~

Considering plane wave propagation along the

x axis, and choosing our coordinate system so

that there is no component of the earth' s

magnetic field along y, cwceatsdenote the x and s components of this 6eld by the symbols IIL, and

II~, respectively.

For Maxwell's field equations (expressed in

Lorentz units), we may write:

Curl

H=

1 /BE +g

(,

~

where E and II are the electric force and mag-

jnetic force of the plane wave. is the convection current density. %e there-

fore have:

j =dP/dh,

where I' is the polarization at any point.
When expanded, Maxwell's equations become:

BE,/Bt+j, = 0,

„), BH./Bx = (—1/c) (BE„/81+j

(5)

BH„/Bx= (1/c) (BE,/Bt+j, ),

(6)

BH,/Bt =0,

—~E*/» = (1/c) (~IIu/~~) — (8)

BE„/Bx= —(1/c) (BH,/Bt).

(9)

~~ E. V. Appleton, J. Inst. Elec. Eng. , wireless section,

Vp '20S'K7 .(1F9o3r2st)e.rling and H. Lassen, Ann. d. Physik 18,

26 (1933).

'06 W. G. Baker and A. L. Green, Radio Research Board

Re'po' rGt .

No.
%V.

3 (Melbourne, Pierce, Cruft

1932). Laboratory

lectures.

PHYSICS OF THE IONOSPHERE

Erl. (4) indicates that the waves are not entirely transverse since there will ordinarily be a longitudinal component of electrical force. Erl. (7) indicates that there is no longitudinal
component of H.
The remaining four equations determine four wave equations. Assuming that the 6eld vectors contain the factor e'~&' ~', these wave equations
simplify to:
(c'q' —l)E„P„0=,—
(c'q' —1)E,—P.= 0,
(c'q' —1)H, cqP„=—0, ' (c'q' —1)H„+cqP, = 0,
whence
I' „/E,=P„/P. and II„/II,= E./E, . (—11)

magnitude of the negative charge of the electron,

and g is a friction factor. Assuming that all dependent variables contain

time only in the factor e'&', these equations

become:

EE,„==((cax++iiPP))PP, +„iiyyrrPP„,—+, ipcP„

(12) (13)

E,= (a+iP)P, inc—P„,

(14)

where

a = —mp'/Ne',

(13)

p =pg/¹',

(16)

¹' mpeHc/mc

(17)

mpeHr/mc

Consequently the resultant components of
magnetic force and electric force are at right
angles to each other in the wave front.
If we now study the motion of an electron by
considering its displacement components, $, g, i,
along the x, y, and z axes, we may write the equations of motion:

d'$ m

=

—eE

—gd—$ eHr

dq

dt'

dt c

— m = -eE„-gd—q —eHc di +eHr d$

dt2

dt c dt c dt

—+ — = — d2 1.

pl

eE g

di' g

eHc d$

dt2

dt c dt

or in terms of the electric polarization:

¹ ¹ ¹ t5 d P~ = —eE,—g dPg IIP dP71

dt2

dt

dt

¹ ¹ ¹ ns d'P„ eB g dPy Hl. dP, +IIg dP„

dt2

dt

dt Nc dt

¹ tpz d2P g = —eEg —gdP, +III. dP,

Ne dt2

dt

¹1; Since P = Neg; P„=Neg; P, =

where N is

the number of electrons per cm' and e is the

In these expressions e, III., and IIz are ex-
pressed in Lorentz units. If e and II are to be
expressed in electrostatic units (for convenience

in computing numerical values for the coefficients) each expression should be divided by 4m.
Lorentz suggestediver the rough approximation, g=2nzv, where m mould be the mass of the
electron in this case, and v would be the average number of collisions per second which the electron makes with the air molecules. The
expression has been accepted without comment
'" by several writers, though its validity is doubt-
ful. For the present it seems better to leave g in the expression as an empirical constant.
From (4) and (12) we obtain:

P*= Pw/(—1+~+iI)

(19)

From (19) and (13) we have:
(20). P„ E„=(n+iP) —(y 'r(/1+a+iP))P„+iyr.P.

E„P„, Combining (20) with (14), and eliminating the
E„, and P, :

x

(~+ip)— 1+a+iP

c'q' —1

(21)

whence

cq =1+—
2(~+'0)

V"/(1+~+'—0) ~L~"/(1+~+'0)'+4& c'3'

(22)

HARRY ROKE M I M NO

For convenience us write:
p02=4wNe'/m,

in analyzing this expression let
pc eH——c/mc, pr eH——r/mc,

where e, III., and II~ will now be expressed in

electrostatic units. Whence
p'/—p', P =pg/po'm,
&&=ppl/po ~ »=pp&/po.
Substituting, we obtain:

c'q'= (p ick/—p)'
2(P,Pg/m)

2pp
(P Pr /(P —P, —iPg/m)) ~LP P,'/(P' P,' —iPg—/m)'+4P'P" j~ (23)

In general, therefore, the ray treatment of the problem indicates that we are to expect two diA'erent indices of refraction, and two corresponding modes of propagation through the
medium. To fInd the polarization belonging to
either of these modes of propagation we may use the auxiliary equation:

H„

1YL

H, 1/(c'q' —1) —(a+i')

(24)

obtained by substituting (10) in (14). If we had included the Lorentz correction
term, we would have obtained the following
equation in place of Eq. (23):

c'q'= (p —ick/p) ' (2p'+ '.p 02—ipg-/m) —p'pr'/(p'

2p0
—3 po' —ipg/m)

~ Lp'pr'/(p'

'po' ipalm—)'-+4p—'p'3' (23')

Disregarding the slight additional modification produced by evaluating the friction factor, g, in terms of the collision rate, v, we may note that
Eq. (23') is essentially the Appleton-Hartree
formula, which has been used extensively in
numerical computations by TaylorIos and a
number of others. For the reasons mentioned in Section E, however, we now prefer Eq. (23) wliich is equivalent to Appleton's"4 simpler
formula.
Equations, similar to those derived above, have been widely used in ionosphere theory and
have been extensively supported by experiments. VVe will consider their signihcance in further detail in several of the following sections. How-
ever, it would be incorrect to assume that the magneto-ionic theory of the ionosphere has received universal acceptance or that it represents the only possible attack on the problem. Before beginning an analysis of the conventional theory, I will therefore insert several references representing viewpoints not covered by the
derivations above. A rival theory oAered by Eckersleyl'9, iso has
received a considerable amount of attention.

110"8 TM.a,Lry.

ETcakyelrosrl,eyP, roJc..InPshty.s.ElSeoc.c.E4n5g,.

245
71,

(1933). 405 {1932).

»0 T. L. Eckersley, Phil. Mag. 9, 225 (1930).

Eckersley contends that magneto-optical effects are only of major importance at night and perhaps on extremely long waves in the daytime. He considers that daylight transmission is primarily controlled by absorption phenomena which in turn depend upon whether the frequency of the radio wave is higher or lower than the average collision rate of the electrons. At an early date Mesny"' contrasted Eckersley's hypothesis with the conventional theory and oA'ered interesting criticisms of the theories of Lassen, Nagoaka, Pedersen, Breit, and Chapman, and the experiments of Pickard.
The validity of the ray method of electric wave
" analysis has been questioned by a number of
writers. Eckersley'" wishes to substitute an approximate phase integral treatment, which presents a formal analogy with quantum theory and uses the methods of Bohr and Sommerfeld. He considers the ray method inadequate for long waves. Hartree~ finds that the conditions for use of the ray method are not completely satisfied, but the correction introduced by the

»~ R. Mesny, Onde Elec. 7, 130 (1928}.

»~ T. L. Eckersley, Proc. Roy. Soc. A132, 83 (1931).

»»~'TT..

L. L.

Eckersley, Eckersley,

Proc. Roy. Soc. A136, 499 (1932). Proc. Roy. Soc. A13V', 158 (1932).

PH YSI CS OF THE IONOSPHERE

"' wave method is unimportant. Epstein'"
justifies the use of geometrical optics since there is no appreciable reflection unless the conditions approximate those of total reflection. He considers that the refracted ray is properly traced by the ray method and this refracted ray carries the main part of the energy.
" Finally, we may refer very briefly to a theory
based on "subelectrons, offered by Bagchi"' and to a theory involving diffusion by little
"' cloudlets of electrons, advanced by Ponte and
Rocard. Apparently neither of these theories has attracted much attention in the general
literature.

H. ANALvsIs QF MAGNETQ-IQNIc
DOUBLE R,EFRACTION

The elementary theory, presented in Section D, indicates that we may expect that a radio wave, directed vertically upward, v ill experience total internal reflection when it arrives at a level where the electron density is just sufhcient to reduce the refractive index to zero. We developed a simple expression relating the refractive index to the electron density and to the angular frequency of the wave.
In Section G we undertook a more thorough investigation of the problem, taking account of
the modifications produced by terrestrial mag-
netism. As a result we obtained a more elaborate formula (Eq. (23)) for the refractive index, involving the longitudinal and tangential components of the earth's magnetic field in addition to
the quantities previously mentioned. Our improved equation indicates that there are two different numerical values of the complex re-
fractive index corresponding to a given electron
density, wave frequency, and direction of propagation. We find, therefore, that the ionosphere is a doubly refractive medium, and we are to expect two modes of propagation of radio waves in the ionized regions.
In general we find that a radio wave, emitted as a single ray from a sending antenna, mill be
resolved into an ordinary and an extraordinary

»'
""''"s

P. S. P. S. S. C.
Ponte

Epstein, Prac. Nat. Acad. Sci. 16, 37 (1930). Epstein, Proc. Nat. head. Sci. 16, 627 (1930). Bagchi, Nature 134, 701 (j.934). and Roeard, Comptes rendus 1ST, 942 (1928).

ray during its passage through the ionized layers. These rays will follow diferent paths, will penetrate to different levels, and will return to earth at diferent times. If both waves happen to return to the same point at approximately the same time, they recombine to produce a resultant signal at the receiving antenna. Under
such conditions the resultant wave frequently
undergoes rapid change in intensity and polar-
ization as a result of minor alterations in the
path of either ray. Under other conditions one
of the component rays may be lost by absorption or by penetration, or it may be delayed so long that we may clearly recognize it as a separate echo when it finally returns.
Until very recently, practically all direct
measurements of the properties of the ionosphere
have been carried on by means of transmitting and receiving stations placed relatively near together, so that the waves have been "reflected"
at normal incidence by the layers. Two main
types of experiments are often carried on under such conditions. In "variable frequency" experi-
ments the angular frequency of the emitted wave is altered over a wide range while the
corresponding echo sequence, obtained at the
receiving point, is recorded automatically or noted by an observer who watches a cathode-ray
" oscillograph screen. In this way it is frequently
possible to obtain a pair of "critical frequencies. One of these "critical frequencies" is the highest frequency at which the extraordhnury ray will be "rejected" by the electron concentration prevailing in a given layer at the time of the experiment. The other "critical frequency" is the
highest frequency which will permit the return
" of the ordinary ray under the same conditions.
Evidently each "critical frequency, measured accurately by direct experiment, is just high enough to reduce to zero the refractive index for the corresponding mode of propagation, at the point in the layer where the electrons are most dense. For higher frequencies the corresponding refractive indices do not decrease to zero at any point in the layer, and one or both of the component rays will penetrate to higher layers or will escape to interstellar space. If we may assume the validity of the analysis presented in Section G, it should be possible to compute the electron density and the strength of the earth's field from

HARRY ROKE M I M NO

our measurements of critical frequency. As the complete experiment occupies but a short time, the electron density in the layer remains sensibly constant during the test.
"Constant frequency" experiments, on the other hand, are usually continuous experiments designed to study the normal and abnormal variations in electron density. In studying the
F region a transmitter frequency is selected
which will normally show no vertical incidence reflections from that layer during the early
morning hours (say 3492.5 kc). During the sunrise, period the electron density increases as a result of the absorption of ultraviolet light.
Eventually the electron density becomes large enough to reduce to zero the refractive index for the extraordinary ray (in the numerical case just mentioned). The exact time of the beginning of
this type of echo may be recorded automatically to the nearest minute. Some thirty minutes later the further increase of electron density suddenly permits the "reflection" of the ordinary ray. Similar eAects occur in reverse order as the electron density slowly decreases during the late evening hours. Considerable variations occur from day to day but by the use of automatic. apparatus it is possible to secure suFficient statistical data for the study of ionization rates and decay rates. Continuous measurements of this sort are also extremely useful in accumulating precise data on the effect on radio propagation of magnetic storms, sunspots, auroral
displays, meteors, weather conditions, and other geophysical phenomena. The two types of experiments are complementary.
In interpreting the data of both types of
experiments it is evident that we shall be
interested in studying the variation of refractive index with angular frequency and with electron density, and will wish to know the type of polarization corresponding to each of the two refractive indices. We shall be especially interested in the conditions which just suf6ce to

reduce the refractive indices to zero. The

equations derived in Section G will therefore be

" "' analyzed from this point of view. Such analyses

have Taylor

b'ee'n

given by and others.

Goldstein, The results

RatcliR'e, obtained by

the various writers are equivalent except for the

quantitative differences which result from the

inclusion or omission of the Lorentz correction

term previously discussed. RateliA'e's treatment

is particularly good, as he gives special attention

to the progressive changes which accompany

alterations in the relative directions of the wave

normal and the earth's magnetic 6eld. The

following outline is suggested by his article,

though the notation is altered to some extent

and the discussion is necessarily greatly ab-

breviated.

"' As differences in terminology have caused some

misunderstanding"'

in regard to the actual

polarization of downcoming radio waves, it is

necessary for us to examine the reference axes

and sign conventions with considerable care.

As in Section G, let us consider a radio wave

propagated in the direction of the positive x axis

(Fig. 7), with its own magnetic field entirely in

the y2 plane. The magnetic fieid of the earth is

represented by a vector in the x2' plane, with a

longitudinal component, IXI., along the positive
x axis, and a tangential component, IIz, along

the positive s axis. In the ionized medium the

wave may be resolved into two parts represented

by the two ellipses in the ys plane. The dotted

II„ ellipse represents a ray with "right-hand"
polarization since its magnetic component,

attains its maximum positive value 90 electrical
degrees in advance of the component II,. The

magnetic vector of this ray therefore rotates in a

clockwise sense when viewed in the direction of

propagation. As the opposite conditions obtain

in the case of the solid ellipse, this represents a

"left-hand" polarization.

In the present section we will neglect collisional

friction and drop a11 terms containing the

friction factor, g. With this simpliheation we

may rewrite Eq. (23) in the form:

1— — cpg2= (p, hack/cp)'=
2(l

4pp /4p2)

(l — 2 (cp p2/442)

&pp2/pp2)

4pT2/~2~ L~ 4/~4+44pc2/ip2(g

4p 2/4p2)27$

"''" J. A. Ratcli8'e, Kireless Engineer 10, 354 (j.933).
T. L. Eckersley, Nature 1M, 398 (1932}.

'~' T. L. Eckersley, Nature 130, 472 (1932}.

PHYSI CS OF THE IONOSPHERE

coo' —4s NP/m, H——L,e/mc,
rur = Hre/mc

Ke mill also»se:

= co~ He/mc,

x —X( X1,

Note that X and Y are ratios, not
identi6ed with the space vari-

= Xy 27cc/coy,

ables, x and y.

8=orientation of II, measured

from the di-

rection of propagation.

At a point where the earth's magnetic field is

approximately 0.5 e.m. u. , the numerical value

of )1 is approximately 200 meters. The corre-

sponding frequency is a resonance frequency

produced by the rotation of the free electrons

about the magnetic lines of force. By analogy

with optical problems it is natural to expect a

considerable difference in the properties of the

medium above and below this critical region.

The resonance is not sharply marked since the

strength of the magnetic 6eld varies appreciably

with altitude and since there will be frictional

damping in the practical case. Let us first
consider the special case, ) =100 meters, since

this will illustrate the typical conditions en-

countered in the transmission radio waves. The corresponding

of high curves

ofrfepq,u' eensc.yX

have been computed and plotted by RatcliEe

(Fig. 8) for three difFerent orientations of the

direction of propagation with respect to the

¹ magnetic 6eld of the earth. Notice that the abscissa, X, is directly propor-
tional to the ionization density The shaded

areas enclose a family of curves which might be

drawn for any of the intermediate orientations

included between the longitudinal and transverse

directions of propagation. %'e may arbitrarily

select the 45 degree curve as a typical representa-

tive of this group. By using the lower sign in

Eq. (23) we obtain the continuous cur~e which

passes through the (0, 1) point. Though the

shape of this curve does depend upon the

strength of the magnetic field, we may note that

H has no effect upon the location of the point

where p' becomes zero. Consequently the name

"ordinary ray" is assigned to the corresponding

mode of propagation. By the use of' the upper

FIG. 7. Polarization of radio waves. *

" sign in Eq. (23) we obtain the curve for the
"extraordinary ray. This curve has one in6nite

point, points,

and crosses the
X=1—I'and

horizontal
X= 1+ F.

axis The

at the two
location of

these zero points evidently depends upon the
magnitude of II but not upon its direction.

By examining Eq. (23), it is possible to show

that the ordinary ray has left-hand polarization

„(0), (H,/H.

X(1; for values of

plane polari-

zation for X= 1; and right-hand polarization for

(1; X&1. The extraordinary ray has right-hand

polarization for X

plane polarization for

X= 1; and left-hand polarization for X & i.

(Note that these statements are based on the
assumption that II has a positive component in

the direction of propagation, as in the case of

downcoming waves in the Northern Hemisphere.

The polarizations are reversed in the Southern

Hemisphere. Both waves are plane polarized at

right angles in the limiting case of completely

transverse propagation. )

Let us now apply these results to the practical

problems of short wave radio transmission and

experimental ionosphere research. A radio wave,

transmitted vertically upward, is divided into an

extraordinary and an ordinary component when

it reaches the ionosphere. These two components

continue to move upward through a region of

increasing electronic density. Eventually they

may reach a level where there are enough

electrons to reduce to zero the refractive index of

~ Fig. 1, Wireless Engineer 10, 355 (1933}.

HARRY ROWE M I M NO

2.0

~ LOR OITUDlR& L

—YR) lOVEISI ', a5'

I.O

~~5%%8+ +Ex 4

FIG. 8. Variation of refractive index with ionic density
(~=2) t

the extraordinary ray (X=1—Y). This ray then

experiences total internal reflection, but the

ordinary ray must rise to a greater height in

search of a region of appreciably greater density,
such that X=1. If the layer is sharply defined

the difference in path may be too small to

measure by ordinary means, and the reflected

signals overlap closely. During the sunrise period

and the late evening period, however, the differ-

ence in path is frequentIy large and the two

echoes are easily resolved. It is not unusual for

an 86 meter vertical incidence extraordinary ray

reflection to persist throughout the night when

no trace of the ordinary ray can be found.

The second zero point of the extraordinary
ray, at X= 1+ Y, probably cannot be detected

experimentally, since practically all of the energy

oXf =th1e—r7a,y

has and

been lost in in collisional

the reflection at friction. Theoreti-

cally, even in the case of "total" internal

reflection, a small amount of energy does pene-

trate into the outer region, but this becomes

inappreciable in a distance of the order of a

wave-length.
X= 1 —I and

Consequently
at X= 1 are

the zero points at
the points which may

be correlated most directly with experiment.

In the case of nonvertical incidence the results

are qualitatively similar. The extraordinary and

ordinary rays rise to different heights, as their

refractive indices must become small, though

they need not be reduced to zero. The electron

density may be sufhcient to return one or both

components to earth at distant points while

f Fig. 9, Wirekaa Engineer 10, 359 (1933).

insufficient to produce reflections of either type
at small angles of incidence.
I.et us next consider the behavior of waves

which are appreciably longer than 200 meters.

Fig. 9 represents Ratcliffe's study of the propa-
gation of 400 meter waves (7=2). Boundary

curves have again been drawn for the limiting

cases of transverse and longitudinal propagation,

with a representative intermediate curve for
8=30, and shaded areas to show the transition

region.

It is apparent that the ordinary and extra-

ordinary rays have exchanged roles, the extra-

ordinary ray now producing the echo which

travels over the longer path and is reflected from

the more dense region. Although the extraordi-

nary ray evidently reverses its polarization as it
passes through the level where X= 1, the

extraordinary ray again has right-hand polariza-

tion in the lower levels of the earth's atmosphere

where it may be examined experimentally. The

ordinary ray retains its left-hand polarization.

(Both statements apply to downcoming waves

in the Northern Hemisphere. )

In conclusion we may refer briefly to a few

representative experiments which illustrate the

methods used in the study of ionospheric double

refraction. The construction of a typical polar-

imeter receiver for studying the nature of

'" "' "' downcoming waves is described by Ratcliffe and

White.

Martyn and Green"4 have used

a three-aerial system of measurement for de-

termining the polarization of downcoming waves

in the Southern Hemisphere. Their results differ

from the results obtained by similar measure-

ments in the Northern Hemisphere and the

differences are correctly predicted by the usual

"' magneto-ionic theory. Berkner and Wells"' have compared critical

frequencies measured in Washington, D. C., and Huancayo, Peru, and conclude that the results

are substantially in agreement with the known

'~ J. A. RatcliEe and E. L. C. White, Phil. Mag. 15,

12"5 '(J1.9A33. )R. atcliHe and E. L. C. White, Phil. Mag. 10,

423»

(1933}.
D. F. Martyn

and A. L. Green,

Proc. Roy. Soc. 148,

"104 (1935).

'

'LA..

L.
V.

Green, Proc. Berkner and

I. R. E. 22, 324 {1934). H. W. Wells, Proc. I. R.

E.

22,

"' 680 (1934).
L. V. Berkner and H. W. Wells, Proc. I. R. E. 22,

1102 (1934).

PH YSICS OF THE IONOSPHERE

difference in magnetic intensity. Appleton"' and Chapman"' consider that it is now possible to make an accurate determination of magnetic intensity as a function of altitude by means of the precise measurement of critical frequencies. Preliminary results are in accord with theory.
Though the high frequency echo component of lesser delay is normally the extraordinary ray, Appleton and Bmlder'~ have found that group retardation in a lower layer can cause a reversal of position when the experimental frequency is just above the penetration frequency of the low layer. This reversal is caused by the fact that the ordinary ray has the higher group velocity.
By employing exceptionally good experimental
technique in oscillograph measurements on dis-
crete pulses, von Handel and Plendl"' made a thorough test of the distortion of a radio signal (selective side-band interference) produced by double refraction in the ionosphere.
Green and Builder"' have explained and interpreted observations of Hollingsworth, Naismith, and Namba on the rotation of the plane
of polarization of long radio waves.

I. COLLISIONAL FRICTION

The effect of collisional friction has been

considered at some length by a number of

writers, but no general agreement has been

reached as yet in support of any single definite and quantitative mathematical treatment. I

shall therefore limit this discussion to a brief

qualitative survey with references to some of the

typical exploratory computations undertaken

by various research groups. The problem may be attacked from a number of different angles.

We may be interested in the nature of the

molecular and atomic collisions, or in the fric-

tional loss which might limit the useful trans-

mitting range, or in the selective absorption which alters the polarization of a "reflected"

'" wave. Hulburt"2

has contributed a number of

interesting papers on the absorption of radio

"' E. V. Appleton, Nature 133, 793 {1934). "' S. Chapman, Nature 133, 908 {1934).

j3~ P. von Handel and H. Plendl, E. N. T. 10, 76 (1933).

si A L Green and G. Builder, Proc. Roy. Soc. 145,

14'5~ '~

{E1.9034. )H. ulburt,
E. O. Hulburt,

Phys. Rev. 29, 365 (1927). Phys. Rev. 29, 706 {1927).

waves in the upper atmosphere. By treating
collisions between electrons and molecules he develops a simple attenuation formula and considers that the measured values may give
data on electronic and molecular densities at
high altitudes. Yokoyama and Nakai'" find that the ob-
served east-west attenuation is decidedly greater than the north-south attenuation in the case of long wave transmission during daylight hours in fairly high latitudes. However, the agreement between experimental measurements and the various theories which have been offered is none too good.
G. Kreutzer'" examines the absorption of a finite wave train in a dielectric and finds that the absorption constant depends on depth of penetration into the medium, on the frequency, and on the length of the train. The absorption is smaller than that given by classical optics theory for frequencies near resonance. The theory has been tested by experiments on ethyl alcohol.
By assuming that the directed momentum of the electron is destroyed at each impact, and that the velocity acquired between collisions is small compared with the random velocity of thermal agitation, Childs"' "~ computes the theoretical conductivity of a gas and finds that it is of the same order of magnitude as the ob-
5.0
—UNOITLS NAL —--- TI IANSVB ISF
Si)4

FIG. 9. Variation of refractive index with ionic density (V=2).f

'3' E. Yokoyama and T. Nakai, Proc. I. R. E. 1'I, 1240

(1929).

"' G. Kreutzer,
" E. C. Childs,

Zeits. Proc.

f. Physik 60, 825 (1930). Phys. Soc. 44, 246 (1932).

'3~ E. C. Childs, Phil, Mag. 13, 873 (1932).

f. Fig. 10, Wireless Engineer 10, 360 (1933).

HARRY ROKE MI MNO

served conductivity. He concludes that the kinetic theory is valid for ionosphere compu tations.
Using the Appleton-Hartree formula in a form similar to that derived in Section G (but with explicit use of the Lorentz polarization term and the Lorentz evaluation of the friction factor in terms of collision frequency), Taylor" computes dispersion curves for four radiofrequencies and four collision frequencies. The insertion of the frictional term removes the infinities from the dispersion curves presented in Section H, but does not affect the general shape of these curves or the conc1usions derived from them. However, an important additional piece of information now appears. (Similar but less complete discussions have been presented previously by other writers. ) At broadcast frequencies and in the Northern Hemisphere the extraordinary ray is greatly weakened by attenuation and absorption. Though the ordinary ray penetrates deeper into the ionosphere, and experiences a longer delay, it eventually returns with a greater amplitude than the extraordinary ray. In the case of downcoming broadcast waves in the Northern Hemisphere we therefore find that left-hand polarization normally predominates. However, if' the electron density drops through a critical value and the weak extraordinary ray alone remains, the polarization changes suddenly to the right-hand type. These changes have been observed experimentally by Appleton and Builder, '~ White"' and others, and this portion of the collisional friction theory appears to be in entire accord with experiment.
From our present viewpoint it is somewhat unf'ortunate that the Lorentz polarization term was used in Taylor's elaborate and arduous numerical computations. Though most of her results are undoubtedly qualitatively correct, the importance of the subject would seem to justify a repetition of the quantitative work in accordance with Darwin's reexamination of the problem. In addition to the polarization eft'ects mentioned above, Taylor finds a transition from quasi transverse to quasilongitudinal transmission at a critical collision frequency. By considenng indices of attenuation in addition to
»1l F. %'. G. Khite, Proc. Phys. Soc. 45, 91 (1934}.

indices of refraction, she also concludes that the lower boundary of the lower layer must be sharp in the optical sense.
"' Io short-Chshmce observations on long wave
phenomena, Naismith"' finds that the strongest downcoming wave is obtained in the northsouth direction, and he therefore believes that
there will be correspondingly /ess energy available for /ong dishzece transmission in this direction. He also finds that magnetic storms increase
the long wave field at short distances but decrease it at great distances.

J. COMPLETE ANALYSIS BY CONFORMAL
REPRESENTATION

Bailey'4' and Martyn'" have recently developed an interesting graphical method for obtaining quantitative solutions of' the Appleton-
Hartree formula (or similar equations for the complex refractive index of the ionosphere), taking full account of the damping produced by collisional friction as well as the effect of the earth's magnetic field.
Though this new attack on the problem
deserves special mention in an individual section
of this report, it does not seem necessary to
introduce any extensive description of the specific geometrical processes involved. The new pro-
"' cedure is a straightforward application of the
methods of conformal mapping, and it is fully described and illustrated in two recent articles"' '4' which are available in practically all scientific libraries. "The two charts needed
for the determination respectively of the polar-
ization and the refractive index are easily drawn,
" as they involve only circles, ellipses, and para-
bolas.
Martyn'4' has carried out the graphical analysis for "five typical wave-lengths (100 m to 20,000 m), three collisional frequencies (10', 10', 10', per sec.) likely to cover the range of practical importance for radio propagation, and for di-
" rections making the three angles 0', 40', and 90'
with the magnetic field. In general his results

"' R. Naismith, J. Inst. Elec. Eng. 69, 875 (1931). '" R. Naismith, %'ireless Engineer 8, 254 (1931).

~"

J. A. Bailey, Phil. Mag. 18, 516 {1934).
J.D. F. Martyn, Phil. Mag. 10, 376 (1935).
H. Jeans, Electricity aed MugeeIism (Cambrirlge,

&908).

PHYSICS OF THE IONOSPHERE

verify and supplement the conclusions presedent
in Sections H and I of this report. In the detailed
analysis of the limiting values of polarization obtained with low electron densities, Martyn's results differ essentially from those of Taylor'8
"' though they are in accordance with the conclu-
sions of Baker and Green. As Taylor and Martyn have studied the same fundamental equation (the Appleton-Hartree formula, including the Lorentz polarization term), the difference is evidently due to a minor analytical discrepancy which can undoubtedly bt' removed. As Martyn refers, in a footnote, to the debatable character of the Lorentz polarization term, we may hope that he will soon discuss the quantitative modifications which would result from its
omission.
K. FINE STRUCTURE OF THE IONOSPHERE
%hen reviewing the basic experimental facts (Section C), we noted the existence of two main "reHecting" regions, commonly designated by
the symbols E and F. By an analysis of the
elfect of the earth's magnetic field (Section H) we were led to expect two modes of propagation in each region. The echo pattern obtained at the receiver is further complicated by the common occurrence of "multiple" reHections, representing waves which have been "reAected" repeatedly from one or more of the layers and from the earth. (In our own measurements we have observed as many as 18 round trips of this chars. cter at times of low attenuation. )
In addition to these well-known effects, however, a number of observers have noted the presence of echoes which can only be accounted
for by assuming that the F layer commonly
divides into two parts during the daytime; and by assuming that additional "reRections" may sometimes be produced by minor concentrations of electrons above, between, or below the main
8 and F regions. Thus echo analysis becomes
almost as complicated as spectrum analysis and special symbols are obviously needed in order to
describe the "6ne structure" of the echo pattern. As the best measurements have been made by a
relatively small number of experimental groups, the eR'ect of geographic differences is not completely understood, and the literature in regard

to "fine structure" is still somewhat contra-
dictory. Further confusion arises from the various terminologies and symbols employed in diferent countries. Nevertheless, references to a few representative experimental reports in regard to
stratification may be of interest. Wherever
applicable I shall use the symbols adopted by international agreement at the London conven-
tion of the U. R. S. I. in 1934, though other
notations are often used in current periodicals.
As early as 1916, while considering the propagation of radio waves in the atmosphere, Lowenstein'~ stated, "Measurements of the intensity of light taken at sunset show three distinct discontinuities, when the last rays
" become tangent to the layers of air at the
height of 11 km, 75 km, and 220 km. As yet we cannot know whether or not Lowenstein's 75 km and 220 km "discontinuities" are closely
8 related to the and F ionic regions, but addi-
tional observations of this character might prove to be valuable.
" Using his original experimental methods
(studying natural "fading, and producing artificial interference changes at the receiving point
"' by slowly varying the frequency of the trans-
mitter), Appleton and his co-workers"' noted the existence of two reflecting" regions, and
assigned to them the symbols E and F. Ap-
parently these letters were chosen in order to
" provide designations which could later be ex-
tended to lower or higher "layers. Appleton
also found evidence of an absorbing layer below
the E layer.
During the same period, Breit, Tuve and DahP' '4' developed the "pulse" method of observation (transmission of a very short radio
signal and measurement of the time 1ag of the echoes), and obtained layer "heights" ranging
from SS to 220 km.
In honor of Appleton's numerous contributions
to ionosphere research, other English writers

"'"4 F. Lowenstein, Proc. I. R. E. 4, 271 {1916).
E. U. Appleton, Tijds. Nederland Radiogenootschap

2, 115 (1925).

~4' E. V. Appleton and M. Barnett, Proc. Roy. Soc. 113,

450 (1926}.

I47 E. V. Appleton, Nature 120, 330 (1927).

E. V. Appleton and A. L. Green, Proc. Roy. Soc. 128

159 (1930). '~' G. Breit,

M. A. Tuve,

and

0. Dahl,

Proc. I. R. E.

16, 1236 (1929).

HARRY ROKE M I M NO

" have often referred to the F layer as the "Apple-
ton layer, reserving the name "Kennelly-
Heaviside layer" for the E layer, which pre-
sumably played a greater part in the transmission of the long waves used in Marconi's first transatlantic experiments. (Kennelly originally postulated more than one refIecting level„while Heaviside referred to a single layer. ) In general, however, the term "Kennelly-Heaviside region"
" has been used as a synonym for the word
"ionosphere, since the use of proper names as designations of the individual parts of the ionosphere seems likely to produce additional
confusion. After the existence of two main layers was
apparently well established, Goubau and Zen-
neck'" published a report ascribing all reRections to a single low layer. Their paper was summarized
'" in English by Howe"' and it was soon followed
by contributions from Eckersley, Schafer and Goodall, '~ Gilliland, Kenrick, and Norton, '~
pointing out the overwhelming experirriental evidence indicative of at least two layers.
Let us now consider some of the evidence in regard to the minor layers previously mentioned. In order to interpret transmission data taken in China in 1927—28, N. H. Edes'" suggested the existence of a low lying region of small ionization. He estimated the height of this layer as 10 km. Recent experiments made by Colwell'" tend to
" verify the existence of such a "C layer.
"' In 1927 and 1928, Appleton ' ' Heising and
Goldstein~ found indications of an absorbing
region somewhat below the E layer. This (partly
" hypothetical) region is now referred to as the
"ozone layer" or "D layer. In general its
"' existence is inferred from indirect evidence,
though Appleton, Goubau" and several others have reported weak reRections. Estimates of its
height range from 30 km to 65 km. Goldstein

'" believes that it occurs at the Lindemann-Dobson
temperature inversion height. Lugeon'" re-

ports a 50 km layer over France. Additional

'" observations have been reported by Bontch-
Bruewitch'" and Sillitoe but Kirby and Judson'" maintain "that if the absence of re-

fractions above 5000 kilocycles during summer

" midday is due to absorption, the absorption is
not mainly below the E layer.

In 1933, Shafer and Goodall'" reported an

"intermediate" assigned to it

layer at
the letter

150 km and
"3L" Their

tentatively observation

'" was soon verified by Appleton'" and by Ratcliffe
and White. "~ This use of the letter "III"

appears somewhat unfortunate, since it breaks

the established alphabetical sequence, and since

the same letter has been used in England as a

graphic description of an entirely different type

of ray path, which presumably follows the zigzag

route illustrated in Fig. 10.
If my interpretation of the recent U. R. S. I. recommendation is correct, the main E layer

should now be designated as E~ while the

"intermediate" layer is to be referred to as E2.

However, there are additional complications
in the E region. In addition to the normal day-

time ionization (presumably caused by simple

ultraviolet absorption), practically all observers
" " have noted the frequent occurrence of "ab-
normal, "sporadic, or "nocturnal" reRectioos

from approximately the same height. To distin-

guish this random, intermittent reflection from
" the regular daytime effect, RatcliAe and White'"
suggest the use of the lower case letter, "e, as

a description of the sporadic phenomenon. In

8" the present review I shall use the designation

"abnormal

to describe this nocturnal re-

Rection.

The critical frequency which is just high

enough to penetrate the E& layer is denoted by

~" G. Goubau and J. Zenneck, Zeits. f. Hochfrequenz-

technik
'" G.

3'F,
W.

200.7H(o1w9e3, 1W). ireless

Engineer

8, 463 {1931).

'" T. L. Eckersley, Marconi Review 31, 1 (1931).

'" J. P. Schafer and %'. M. Goodall, Proc. I. R. E. 19,

'" P14ro'3~c4.T{I.1.9RR3..1GE).i.ll2il0a,nd2,86G{.1W93.2K).enrick, and K. A. Norton,
N. H. Edes and J. C. Coe, Proc. I. R. E. 20, 740

(1932).

"'»~
''

R. E.

A. V.

Heising, Appleton,

Proc. I. R. E. 15,
Proc. Roy. Soc.

75 {1928). 126, 542 (1930).

G. Goubau, E. N. T. 10, 72 (1933).

'~9 J. Lugeon, Comptes rendus 191, 525 {1930).

"' J. Lugeon, Comptes rendus 191, 676 (1930).

'" M. A. Bontch-Bruewitch, Nature 133, 175 (1934).

'"'"SS,. SSil.liKtoier,byCaann.dJ.ER.esBea.rcJhuds1o1n,,

163 (1934).
Proc. I. R.

E.

23,

'" 733 {1935).
J. P. Schafer and W. M. Goodall, Nature 131, 804

'" (1~93"3E).. U. Appleton, Nature 131, 872 (1933).
J. A. Ratclifk and E. L. C. White, Nature 131, 873

{1933).

~67 J. A. RatcliEe and E. L. C. White, Proc. Phys. Soc,

46, 107 {1934).

PHYSICS OF THE IONOSPHERE

FIG. 10. "M" reAection.
the symbol fs, If .the resolving power of the
experimental apparatus is suf6ciently high, the
two component critical frequencies, f'E, and f's,
corresponding to the ordinary and extraordinary
ray™ybe separated. (Magneto-ionic splitting
of the Ir" layer has been observed in our own experiments, though the eR'ect is much more
conspicuous in the F region on account of the
smaller electron-density gradients encountered
at the higher levels. )
Similar indications of substratification have
been found in the F region. Kirby, Berkner, and
Stuart'" note the existence of an Fi layer
somewhat below the main F~ layer. The boundary becomes indistinct in winter and at night, but is frequently distinguishable during summer daylight hours. Appleton'" confirms this observation. Having in mind the graph of ion density as a function of altitude, he considers that FI is a sort of ledge or protuberance on the
lower side of the main Fg region. It is quite
possible for the two types of reflection to appear simultaneously. In fact Henderson"0 calls attention to the occurrence of anomalous echoes from the high levels, which occur at times when
Region E is much more intensely ionized than
region F.
As the experimental frequency is progressively increased, the F~ region becomes completely penetrable, and eventually the F~ reflections also disappear. In the latter case, however, we do not have an entirely clear indication of the sudden decrease in group velocity which definitely marks the penetration frequencies of the lower layers. The reflections often. fade out somewhat
"8 S. S. Kirby, L. V. Berkner, and D. M. Stuart, Proc. I. R. E. 21, 757 (1933).
6 E. V. Appleton, Electrician 11O, 857 (1933).
J. ~~o T, Henderson, Proc, I. R. E, 22, 679 (1934),

gradually without a marked increase in the
equivalent height. Furthermore, as F2 is presumably the outermost strongly reflecting shell, we cannot demonstrate its penetration by exhibiting powerful echoes from points beyond. The indirect evidence afforded by skip zone observations appears to be entirely consistent with the common assumption that electron limitation determines the short wave limit of ionosphere transmission and that waves of very
'" high frequency therefore escape into inter-
stellar space. However, several writers" have suggested that the very short waves merely pass so far into the F2 region that they are completely absorbed. Our own measurements support the "complete penetration" hypothesis, but the question cannot be regarded as definitely settled.
Magneto-ionic double refraction effects may
be observed without difficulty in the F region. Consequently the symbols F Fo'&, &, F'2, and F &
are needed in order to describe the complete set of first order reflections. The experimental results
are in entire agreement with the magneto-ionic
theory presented in Sections H and I. Several
types of polarimeters have been employed. Some
0. statistical results will be presented in Section
Although no strong and steady reflections from
points more distant than Fl are commonly observed, there are many indications of echoes which are weak, or intermittent, or diffuse, or occasional. In our earliest measurements, made
with simple mechanical oscillographs, strong echoes were frequently received from points more that 1500 km distant, but these echoes
ordinarily vanished in a fraction of a second. Similar effects may be readily demonstrated by
'" projecting the echo pattern on a cathode-ray
oscillograph screen. Taylor and Young'" and
" Quackand Mogel"4 '"noticed anomalousechoes
at points within the normal "skip zone, and it is probable that these are due to the "scattering" phenomena mentioned in Section R. By
operating our own continuous automatic re-

'"'~

T.
A.

L. Eckersley, Wireless H. Taylor and L. C.

Engineer Young,

P4ro, c2.13I.{R19.2E7.).16,

561 (1928).

"~A. H. Taylor and L. C. Young, Proc. I. R. E. 17,

'" 1491 (1929).

'"'EE..

Quack Quack

and and

H. Mogel, H. Mogel,

PEro.cN. .I.TR. .6E, N. 1(F1,982249)(.1929).

30

HA RR Y ROKE M I M NO

'" corders"~ under maximum sensitivity condi-
tions we have found two di6erent types of extremely weak but systematic reRections'" coming from points distant 600—j.800 km from the transmitter. These echoes sometimes remain
comparatively steady in signal strength and position for several hours. The direction of
arrival of such echoes is not yet known. Hollingworth"' ' ' Finds indirect evidence sug-
gesting that 30 meter waves are occasionally
trapped between the E and F regions and
transmitted long distances by this mode of
'" propagation. Similar ideas have been advanced
by Janco. In order to determine the relation between
"apparent height" and "true height" in the
region we shall probably need improved
simultaneous measurements employing diRerent
ray paths. Suggestions in regard to the "true height" have been made by Schelleng, '~ Kenrick
and Jen, '~ Ranzi'" and numerous others.
In 1927 M. Hals observed strong echoes which
were apparently reflected from regions far outside of the earth's atmosphere, since the returning
signal could be identi6ed by ear, the time lag measured by an ordinary stopwatch. His ob-
'" '" servations were repeated and conFirmed by
Stormer'" and Van der Pol. Stormer considers that these rejections are due to distant clouds of moving electrons, which form a converging mirrnr of toroidal shape as a result of
their motion in the presence of the earth' s
magnetic Field. Van der Pol claims that the enormous time 1ag (of 30 seconds or more)
merely results from an abnormally low group velocity and believes that the reflection takes

»6 H. R. Mimno and P. H. Kang, Phys. Rev. 41, 39$

'" {&932). H. R. Mirnno

and

P. H. Kang,

Proc. I. R. E. 21,

'" 529 (1933). H. R. Mimno and P. H. Wang, Proc. Fifth Pacific

Science Congress (1934), p. 2195,

H. R. Mimno, Nature 13~, 63 (1934).

'"'8~
»'
»~

J. Hollingworth, J. Hollingsworth,
N. Janco, Proc.

J. Inst. Elec. Eng. 72, 229 (1933).
Wireless Engineer 10, 89 (1933}.
I. R. E. 22, 923 (1934).

'~

J.
G,

C. Schelleng,
%. Kenrick

Proc. I.
and C.

R. E. 15, 1471 K. Jen, Proc.

(1928).
I. R. E.

17,

711'(I1.9R29an).zi, N. Cimento 10, 21 (1933).

'8' C. Stormer,
C Storxnerf

Naturwiss. 17, 643 (1929).
Comptes rendus 180, 365 (1929).

'" C. Stormer, Comptes rendus 190, 106 (1930).

'" C. Stormer, Proc. Roy. Soc. 50, 187 (1930).

'" B. Van der Pol, Nature 122, 878 (2928}.

place in the ionosphere. Pedersen'" disputes this theory, since abnormally low group velocities should be accompanied by excessive at tenuation. Though the original observations seem entirely trustworthy, recent attempts to reproduce the echoes have been unsuccessful. Probably they can only be heard during an especially favorable part of the sunspot cycle.

L. WIIY DoEs STRATIFIcATIQN ExIsT?

From a knowledge of the nature and intensity of the solar radiation, and a knowledge of the

composition, the absorption coefhcients, and the motion of the earth's atmosphere, it would seem

possible to give a complete interpretation of the

'" stratification observed in the ionosphere. Un-

fortunately, as Chapman'"

has recently

pointed out, the present tables of atmospheric

pressure, temperature, density, and composition, for heights much above 30 km, are largely

speculative. At best they merely illustrate the

conclusions to be drawn from alternative

hypotheses.

Consequently, the extensive new quantitative

data derived from ionosphere measurements

must play an important part in extending our knowledge of the atmosphere and the incident radiation. Do atmospheric winds exist at great

heights? What is the degree of dissoriation of

the atmospheric constituents? Does hydrogen
predominate at great heights or is it almost completely absent? To what extent is ozone carried by winds? A complete analysis of such questions ~ould extend through the entire field of meteorology. In the present space it is, therefore, impossible to do more than to mention a limited number of references which may be
used as starting points in investigating the theoretical background of ionospheric stratifica-

tion. Current papers are largely contradictory, though often useful in suggesting new experi-

ments. Chapman'" has oHfered 3 theoretical analysis

which covers:

'"'9~
'~
~'~

P. S. S. S.

Q. Pedersen, Proc. I. R. E. 17, 1750 (1929).

Chapman,
Chapman, Chapman,

JUP.r.RoRco.y. P.ShM.yIse..t,eLoSroooncldo.og4n.3,SC2oo6cn.g(51re09s,3s112)(7.19(13943).4}.

PHYSICS OF THE IONOSPHERE

i. Absorption of ionizing radiation
2. Absorption of nonionizing radiation such as that which
produces ozone.
3. Treatment of dissociating radiation where products of dissociation recombine according to the simple law:
dn jdf = I'-an'.

He derives values for the density of the dissocia-

tion products as a function of height, time of

day, latitude, and season. Appleton'95 finds that the measured values of ionization density show

diurnal variations resembling those which Chap-

man deduced in his treatment of ionization

caused by monochromatic radiation.

'" In the course of a theoretical survey Hul-

burt'"

compares radio determinations of

electron density with values obtained from

magnetic theories and other available informa-

tion. He also discusses the significance of the

measured values of the "recombination" rate in
the E and F regions, maintaining that ionic recombination takes place in the B region. He

believes that the decrease of free electron density

in the Ii& region may be ascribed to the attachment of electrons to oxygen molecules. He
considers that the F2 region is due to ionic

winds Bowing outward from the heated regions near the earth's equator.

However, Appleton" claims that Hulburt has

confused the main I'~ region which was dis-

covered in 1926, with the subsidiary shelf, I'~, which w as discovered in 1933. He further

maintains that Hulburt is incorrect in considering
that the E region consists mainly of ions of
molecular mass. Chapman'" finds that true

recombination is more important than the

attachment of electrons to neutral particles
(in the F region as well as in the E region).
Eckersley'" also presents numerical values sup-

porting this viewpoint.
Lassen' ' states that the effect of free electrons

is unimportant in ions, while Conroy'

comparison
' 6nds no

with hydrogen
hydrogen at any

height. He coordinates the data obtained from

'"F96

E. V. E. O.

Appleton, Hulburt,

PNroactu. rIe.

12'F, 197
R. E. 18,

(1931}. 1231 (1930).

'" E. O. Hulburt„Phys. Rev. 45, 822 (1934).

ios E. V. Appleton, Phys. Rev. O'F, 89 (1935).
'" S. Chapman, Proc. Roy. Soc. 141, 697 (1933). "'T. L. Eckersley, Proc. Roy. Soc. 141, 697 (1933).

»~
'~

H. C.

Lassen, E.
C, Conroy,

N. T. 4,
Science

174 (1927). F4, 113 (1931).

observations of meteors, auroral spectra, and the refraction of sound waves in the stratosphere. The dissociation of oxygen has been treated extensively by Kriitschkow. '03 Nagaoka'04 believes that the upper layer is due to the ionization of helium.
Nagaoka also ascribes certain differences in
east-west and west-east transmission to an
'" "' asymmetry in the layer. The matter has also
been investigated by Nakai. Namba"' oHers a general theory of ionosphere propagation and attempts to compute the effect of the altitude of the sun on the properties of the layer.
Various hypotheses regarding the nature of the incident radiation have been examined by Chapman"' in his extensive theoretical treatment of magnetic storms. Elias"' supposes a
permanent ionization produced by corpuscular rays from the sun, plus a temporary daytime ionization due to solar wave radiation. Swann"' notes that the earth's electric charge (as indicated by measurements up to a height of Jo km) would decrease 90 percent in ten minutes if not
replenished in some unknown manner. He postulates a possible slow death of positive charge on the earth and discusses difticulties
with the solar electron-stream hypothesis. He proposes a new scheme of electrodynamics. However, various other writers believe that the earth's electric charge is maintained in equilibrium by constant thunderstorm activity in the great static-producing areas of the earth' s surface.

M. TIDAL EFFECTS IN THE IONOSPHERE
By measuring the signal strength produced by distant radio stations, several observers have obtained indirect evidence of a lunar tidal efkct

~~'S. Krutschkow, J. Applied Physics, Moscow 'F, 61

{1'9~3H0).. Nagaoka, Radio Research Japan, Report 1, 1

(1931}.

~0~ T. Nakai, Electrot. Laborat. Tokyo, Japan, Re-

searches, No. 241 (1928).
"' S. Namba, Proc. I. R. E. 21, 238 (1933). "' S. Namba, J. Inst. Elec. Eng. , Japan 52, 103 (1933}.
"'S. Namba, Electrot. Laborat. Tokyo, Japan, Re-

sea'rc'She.s,

No. 336 (1932). Namba, Radio Research

Japan,

Report

2, 303

"(1932).

'» »'

S. Chapman, Terr. Mag. 35, 77 (1931).

%G.'.EFl.iaGs,.

Zeits. Swann,

f. Hochfrequenztechnik
J.Am. Inst. Elec. Eng.

2'F, O' F,

66 209

(1926). (1928).

HARRY ROKE MIMNO

'" acting on the ionosphere. Stetson"' has
recently reviewed several experiments and described his own observations. Apparently the signal intensity is decreased as the moon passes over the observer's meridian. The cause of the tidal e6'ect and the nature of the ionosphere disturbance remain somewhat speculative. The simple gravitational atmospheric tide is very
small. Using Gilliland's'" experimental data, Vree-
land"' finds an indication of a iednction jn layer height at the time of new moon. However, the layer height observations do not appear to be extensive enough to allow us to discriminate with certainty between the lunar cycle and the period of sunspot rotation, with its associated
magnetic changes.
Breckel"' suggests that Stetson's conclusions apply only to relatively long waves, and considers that the eR'ect is reversed in the 3500-4000 kc region of the radio spectrum. Various addi-
" "' "' tional observations have been contributed by
Pickard, Stoye, Vincent'"-' and Shannon, but no general agreement has yet been reached, and it is obvious that direct layer-height measurements extending over a long period are
needed.
Stetson'" has also studied certain systematic
discrepancies between radio time signals transmitted from the Greenwich observatory and the time signals transmitted from the Naval Ob-
servatory at Washington. He finds that these
discrepancies depend upon the hour angle of the
moon and attain values as large as &0.03 sec.
This alteration is provisionally assumed to be due to a tidal distortion of the earth's crust which may alter the distance between the two observa-
tories, by as much as +32 ft. This distortion is
somewhat unexpectedly large. From the same numerical data he computes the time of transmission of a radio signal across the Atlantic and

"'"4

H. H.

T. Stetson, T. Stetson,

Phys. Earth,

Rev. 3F, 1021 (1931). Bucko, aM the Stars (McGraw-

Hill, 1934).

"' T. R. Gilliland, "' F. K. Vreeland,

Proc. Proc.

I. I.

R. R.

E. E.

19, 19,

114 (1931). 1500 (1931).

"' H. F. Breckel, Radio Engineering 11, 19 (1931).

2' G. W. Pickard, Bull. National Research Council

"' (1931),p. 125.
K. Stoye, Funktechnische Monatshefte (April, 1933),

p.

~~1''5DP2...

Vincent, Shannon,

Onde flee. 5, 554
Wireless Engineer

(1926). 3, 429

(1926).

finds that it is approximately 0.04 sec. This time of transmission also shows systematic variations of the order of &0.01 sec., which are tentatively ascribed to tidal effects in the ionosphere. Though the apparent correlation of time signal discrepancies with the lunar hour angle is of undoubted interest, I feel that the provisional
numerical analysis is somewhat weakened by the
basic assumption "that the radio wave passes
" either way across the Atlantic in equal time
intervals. As the results imply that the effective ground-level speed of the signal is less than half of the velocity of light, it is evident that the properties of the transmitting medium must receive thorough consideration, and it is possible that the west-east and east-west asymmetries, previously mentioned, may invalidate the basic assumption. Furthermore, the paths of the useful rays would probably depend in part upon the vertical-plane directional characteristics of the
transmitting and receiving antennas. As the time of transmission is apparently of the same order of magnitude as the observed discrepancies,
it seems somewhat hazardous to treat the symmetry condition as a self-evident fact.

N. SUNSPOTS, MAGNETIC INDICES) AND
AURoRAL DIsPLAYs

The existence of a close connection between

sunspots, magnetic disturbances, auroral displays

and radio reception is universally admitted,

though the details of the connecting mechanism

are little understood. The technical literature

covering this phase of radio transmission is par-

ticularly extensive.

A large amount of data on the field strength

"' "' "' of distant radio stations has been compiled by

Austin,

Pickard"'

and many others.

When plotted on a yearly, monthly, or weekly

basis the correlation of radio reception with

Wolfer sunspot numbers is obvious. In general,

transmission becomes worse as the sunspot

numbers increase, but the magnitude and even

the direction of the change depends to some

I 222 M3 L

W. Austin, W. Austin

Proc. I. R. E. 15, 825 and I. Wymore, Proc.

(1927).
I. R. E.

16,

166

(1928). n4 G W. Pickard, 285 G W. Pickard,

Proc. Proc.

II..

R. R.

E. E.

15, 15,

83 (1927). 749 (1927).

226 G W. Pickard, Proc. I. R. E. 15, 1004 (1927).

%7 G W. Pickard, Proc. I. R. E. 19, 353 (1931).

PHYSICS OF THE IONOSPH ERE

extent on the wave-lengths employed. In short wave transmission between definite geographic points, Plendl" and Moge122' hnd that the optimum wave-length may be increased as much as 30 percent by the decrease of average ioniza-
"' tion density apparently occurring at a sunspot
minimum. Austin, Yokoyama, and Nakai"' find a closer solar correlation for short waves than for long waves, though Abbot"' believes that seven periodicities in the values of the solar
constant also appear in long wave propagation data.
However, there does noI, appear to be a good day-Io-day correlation of radio field strength with the passage of definite sunspot groups across the central portion of the sun, though the radio data
does follow the daily variations in the magnetic
field of the earth. It is entirely possible that the
sunspots are not the direct cause of terrestrial
disturbances but merely symptoms of some
underlying solar perturbation. If corpuscular rays cause the ionization changes it is also possible that these rays are deviated from the
simple radial path after emission from the sun. Surveys of the earlier experiments on sunspot
correlations have been given by Mesny"' and
'" Stetson. Recent observations on transatlantic "' signals have been presented by Judson. Gunn"'
and Larmor"' have considered the magnetic fields of sunspots, while Dauvillier"' has examined the possibility of a deformation of the
ionosphere under pressure of solar radiation. Though easily obtainable with a minimum of
apparatus, radio field strength measurements present no direct indication of the ionization density at various points in the ionosphere. Depending upon the wave-lengths employed in the test and the geographic locations of the transmitter and receiver, some signals become stronger

"'' H. Plendl, Proc. I. R. E. 20, 520 (1932). H. Mogel, Telefunken Zeits. 13, 32 (1932).

~'0
'»

L.
E.

W. Austin, Yokoyama

Proc. I. R. E. 20, 280 (1932). and T. Nakai, Proc. I. R. E. 19, 882

"' (1''9»3C1R)... "4 H.

G. Abbot, Science 'N, Mesny, Onde Elec. S,
T. Stetson, J. Frank.

60tI'
103 Inst,

(1932). (1929). 210, 403

{1930}.

E. B. Judson, Proc. I. R. E. 21, 1354 (1933).

"''3' R. Gunn, Terr. Mag. 34, 154 {1929).
J. Larmor, Nat. Roy. Astron. Soc. Monthly 94, 469

(1934).

~' A. Dauvillier, U. R. S. I., London Congress (1934}.

and others weaker as the magnetic conditions

and the sunspot numbers change.

During recent years, therefore, we have begun

to make continuous daily records of the actual

reHections received from various levels of the

ionosphere. Such an experiment requires elabo-

rate automatic apparatus, but it yields direct

information in regard to ionization densities.

Ke have now obtained enough reAection records

to permit a preliminary comparison with avail™

able magnetic data. An 86 meter signal, trans-

mitted vertically upward, usually penetrates the
E layer without producing a reHection, but at

times the ionization in this region exceeds the

critical value and the corresponding echo is

photographically recorded. On days when the
average ionization of the E region is high we

should expect to record such echoes during a

relatively large fraction of the 24-hour period.

In I'ig. 11 we have plotted the total duration of

these E reAections in hours per day. In drawing

these curves we have arbitrarily designated
86-meter E layer reAections as "abnormal" when they commence between 9 r.M. and ground level

sunrise. Evidently these "abnormal" reHections

are not caused by simple uniform ionization

arising from the absorption of ultraviolet light.
" All other E layer reHections are described as
"normal. The "normal" reHection curve shows

a decided agreement with the magnetic index

(which has been plotted downward since the

correlation is the correlation

inverse). The coefficient is

numerical value of
—37. Large values

of the magnetic index indicate days charac-

terized by extensive Huctuations of the earth' s

magnetic field. These indices represent average

values obtained from the data of 48 magnetic

observatories by G. van Dijk. They are published

The. in Caroc&e Magnetique des Jours and in Terres-
trial Magnetism and Atmospheric Electricity

"abnormal" nighttime reAections do not show

such a pronounced conditions. (The

correlation with the numerical coefficient

magnetic
is —15.)

In general it appears that there is a reduced

probability of daytime E layer reAections when

the magnetic conditions are disturbed, This

indicates a decreased average ionization under

such conditions or an increased absorption. The

first hypothesis is strongly supported by the F&

layer observations presented in Fig. 12.

HARRY ROWE M I M NO

Whenever the average ionization is low we should expect that morning sunlight would have to act upon the I' region for a relatively long period before producing an ion density great
enough to reHect 86 meter waves at normal incidence. The commencement of reHections
during the sunrise period should therefore be delayed. Under similar conditions the steady decrease of ionization after sunset should termi-
nate the I'& layer reHections at a relatively early
hour in the evening. . The "critical times" at
which I'& reHections commence and cease may usually be determined precisely for the extraordinary ray and for the ordinary ray. During these critical periods rapid changes of group velocity pmduce corresponding alterations in the observed equivalent height of the reflector and it is therefore convenient to speak of a "rising" I'& layer in the evening hours and a "falling" Ii& layer during the sunrise period. The critical times correspond to the attainment of definite critical densities in the ionized region.
With the method of graph plotting adopted in Fig. 12 all four upper curves should rise whenever the mean ionization is low. Apparently this occurs when the magnetic condition is disturbed. To show the general trend we have plotted the ionosphere data and the magnetic index on the basis of ten-day averages. Similar correlations may be obtained by plotting the daily values directly, but some confusion is caused by excessive detail unless the horizontal axis of the graph is greatly extended.
From Fig. 11 and Fig. 12 we have two separate statistical indications of a decrease in average atmospheric ionization on days when magnetic Huctuations occur. This raises interesting theoretical pmblems. Evidently we cannot accept the naive idea of great clouds of electrons, ejected violently from the sun during sunspot eruptions, which might cause auroral displays, magnetic Huctuations, and immediate increases of atmospheric ionization. Furthermore this detailed analysis of day-to-day alterations must even-
"' tually be reconciled with accepted individual
observations'" which indicate that the yearly average ionization is greatest at the maximum
'3' E. O. Hulburt, Physics 4, 196 (1933).

of the sunspot cycle. In order to obtain further experimerj tal information, it would seem desirable to continue the registration of echo patterns over an extended period. Automatic records totaling approximately 10,000 hours are available at present in our 61es, but these cover a limited portion of the solar cycle.
There appears to be general agreement among experimentalists in regard to the eRect of polar aurora on radio transmission in Northern lati-
"' "' "' tudes. Essentially similar reports have been
made by Wagner, Helbmnner, Ogilvie, Sutton'4' and many others. A marked decrease of signal strength occurs simultaneously with the onset of a nearby visible aurora. This eRect is particularly marked in the case of short wave transmission, and is frequently severe enough to shut oR the signa;ls completely. Echoes disappear in a manner which suggests complete absorption. At more distant points less violent effects of the same general nature occur, though
there is some indication of an appreciable time lag between the aurora and the resultant weakening of signals. The visible aurora is accompanied by simultaneous rapid Huctuations in the magnetic field of the earth. Diill'4' states that it is preceded by rapid variation in the direction of
"' arrival and the intensity of radio signals. Dauvi ilier"' divides auroral phenomena
into two distinct stages. He states that an initial cosmic eRect, which is frequently of short duration, is followed by a relatively slow spread of phosphorescence due to excitation, ionization, and the production of ozone. He observes luminous clouds of ions in rapid motion at an approximate height of 200 km and believes that this ionic "wind" is due to strong electromagnetic forces. "Wind velocities" of the order of tens of kilometers per second occur.
"' The nature of the fundamental cosmic eRect
remains somewhat obscure, though Stormer"'
has shown that the spectacular multiple folds of

'4' K. W. Wagner, E. N. T. 11, 37 (1934).

'4' P. Helbronner, Comptes rendus 191, 536 (1930).

24' N. J. Ogilvie, Am. Geophys. Union, Trans. , 14th

Meeting (April, 1933), p. 44.

243 W. Sutton, Q. S. T. 10, p. 23, October, 1926.

'44 D. Dull, Electronics 5, 268 (1932).

'4~
2"

A, A.

Dauvillier, Dauvillier,

J. de phys. et rad. 5, 398 (1934).
Comptes rendus 194, 192 (1932).

'47 C. Stormer, Terr. Mag. 35, 193 (1930).

~48 C. Stormer, Terr. Mag. 36, 133 (1931).

PHYSI CS OF TH E IONOSPHERE

FIG. 1 1 ~

. ..+& &~ L/a~-0(. 0 ASIIORMA

E REFI, FCTIOIg

h, ~«

R/UV Mnn/'t/«

2C

NOI)MAL E REFLE(TION

IR HO jRS Pl R'. OA'I

Ri,A

It,

„

,

R O'S ''/t

~TS OR SHI I RS
, 1l
/'Jht(i

. U, . A. ,.fht, s
,/

. n n~t k/rJ

~
k~

I5

/

V

Y

r r~L r3 .

/

K

t«V'Jlt

'NI

II

J/t

1.

. /'

. U (MAY

.~i n re r

''

/t'ih
~(l II

MA~)NETIIl INQ„X

2C

5

IO

I5 20 25 30 5

IO I5 20 25 301 5

IO IS 20 23 25 30 5

IO I5 20 25 30 I 5 IO 13

JULY
t833

AUGUST

APRIL

MAY
'934

JUNE

LAYFR ACTIVITY AND MAGNETIC INDEX

RISING F LAYER 2 ORDINARY RAY FF HOURS AFTER
„, i~~~ RISING F LAYER
..„„„"w;;,r I

Fj:6. 12.

2 FALLING F LAYER
ORQINARY RAY
HOURS AF TER SUNRSE

/,
l FALLING F LAYER
Qt TRAORQNYIARY
RAY
HOURS AF TER
0— SUNRISE

OJ~ MAGNETIC
0,6
INQEX
0
FEBRUARY

. «n
/ks

1

/~, p,

&

R,5e

/ xi»5/VX vug/3 g~ /5' 4!

JUNE
I933

AUGUST

CRl TICAL F LAYFR TIMES

OCTOBER

I
DECEMBER

FEBRUARY

AND MAGNETIC ACTIVITY

APRIL
1934

the auroral "draperies" may be reasonably

"' explained by considering families of trajectories
of incoming electrons. BrOche"' — has imitated

some of the auroral phenomena by demonstrating Stormer's electron ray hypotheses on a

24' E. Briiehe,

s'
2"

E. E.

Briiche, BrCiche,

Zeits. Zeits.
Zeits.

f. Astrophys. 2, 30 (1931). f. teehn. Physik 13, 336 {1932). f. Physik 54, 186 (1930).

small scale model of the earth. His experimental technique is especially interesting. On the basis of Stormer's auroral theory and the associated theory of long delay radio echoes. Dostal'" has computed the density of the electronic space charge in the incident electron ray, and main-
'2 H. Dostal, Ann. d. Physik 14, 971 {1932).

HARRY ROKE M I M NO

tains that it is insufficient to cause direct reAection of 30-meter radio waves.
Vegard'~ has recently reviewed extensive investigations of the auroral spectrum. He concludes that the electric rays producing the aurora may possibly be regarded as a mixture of electrons and ordinary matter which largely exists in the form of positive ions, precipitated towards the earth from the corona of the sun.
It has also been suggested'~ that coronal matter
to a large extent consists of oxygen. If the
average charge of the ray bundle is negative and numerical?y small, "we may explain the great
" angular distance between the auroral zone and
the magnetic axis point. Larmor'" has also presented speculations on
the cause of the aurora and its effect upon
terrestrial ionization. He observes that the "fact that a quite small density of ions entirely upsets
the optical elasticity of space as regards long
waves, provides a cause preserving ionized
gaseous clouds of astronomical 'size, for example
" in the interior of stars, from rapid dissipation or
dispersal in bulk.

O. MAGNETIc STQRMs AND METEoRIc SHo&ERs
Most of the day-to-day variations in the magnetic index represent alterations in the character and duration of long continued minor fIuctuations in the magnetic field of the earth. Occasionally, however, the earth experiences a
" sudden disturbance of such violet intensity that
it deserves the name of "magnetic storm. In
extreme cases, magnetic storms completely disrupt wire communication as well as radio communication. Fortunately these intense disturbances are usually brief.
Magnetic storms have been known to recur,
"' after approximate 27-day intervals, on as many
as eight successive occasions. Though it is commonly believed that this is due to the rotation of the sun, it is not certain that a storm is associated with a definite visible'sunspot, or other visible form of solar abnormality. In

2~3
~" ~"

L. T.
J.

Vegard, Geofysiske Publ.
L. De Bruin, Naturwiss.
Larmor, Nature 133, 221

9, No. 20, 269 (1934).

11, Oslo (1932).

(1932}.

J. ~~' G. Angenheister and

Bartels, Hand buck der

Experimenters Physik, Vol. 25, Part 1 (Kien-Harms, 1928),

p. 674.

certain individual instances storms of extreme
violence have accompanied definite sunspot groups of unusual size. Skellett~57 offers pre-
".liminary statistical evidence suggestive that
. . the presence of an area whose activity
may be seen with a spectrohelioscope is a
" necessary though not a sufficient condition for a
radio disturbance. . . . There is also some
tentative indication of a definite time lag, of the order of one day, between the transit of the active solar area and the appearance of concomitant terrestrial effects. More complete solar data is needed.
Dellinger" has recently described a "cosmic phenomenon" which caused brief interruptions of long distance, short wave communication on four occasions in 1935. Such interruptions were effective over the entire illuminated half of the globe and were spaced approximately 54 days apart (twice the period of rotation of the central portion of the sun). We were fortunate enough to obtain actual layer height records'5' during extremely turbulent periods associated with three
successive minor magnetic storms. It appears
quite certain"' that such turbulence would be sufficient to produce the effects described by
Dellinger. Until the apparent double period of 54 days is further substantiated, however, it seems preferable to assume that brief intervening
disturbances at the 27-day points may have
escaped observation. It would seem that this
might easily occur if the most active or most
susceptible radio channels happened to lie on the dark side of the earth at the crucial moment.
A number of observers have reported moderate indications of ionization changes associated with meteoric showers. However, the existing statistical data are not sufficiently conclusive and
some of the apparent results of individual experiments might be ascribed to chance. Some astronomers believe that the energy of the heaviest known showers is insufficient to produce an observable ionization change unless we may assume that the visible meteors are accompanied
by very large amounts of meteoric dust. In Fig. 11

'" A. M. Skellett, Proc. I. R. E. 23, 1361 (1935).

ass

..HR. .DMelilminngoer,

Science 82, 351
and P. H. Kang,

(1935).
Phys.

Rev.

43,

769

(1~93"3H}.. R. Mimno, Science 82, 516 (1935).

PHYSICS OF THE IONOSPHERE

we have indicated the time of occurrence of
some of the periodic meteor showers, but no evident correlation appears as yet in our data.
Tentative correlations with commercial trans-
'" '" "' mission records have been reported by Na-
goaka Quack and Pickard. '68 Skel1ett'~
finds a reduction in the height of the E layer
and offers ionization computations to support
his viewpoint. Minohara and Ito'" claim that
meteors greatly increase the number of echoes. Schafer and GoodalP" are somewhat more conservative, but they also find some reason to believe that meteors cause an observable increase in ionization. Interesting theoretical
" papers have been contributed by Lindemann,
'" '" Maris, '-" Mascart, Millman~" and Malzer.
Visual observation of the motion of meteor
trails have indicated the existence of winds at
great heights in the upper atmosphere. This
'" phase of the subject has been discussed by
Hulburt.

P. THUNDERsTQRMs AND BARoMETRIc EFFEcTs

Most of the meteorological readjustments which determine our daily weather conditions take place in the lower 10 kilometers of the earth's atmosphere. Even under extreme conditions the observed ionization in this region is relatively small and one would not expect significant absorption or ionic refraction of radio waves. Nevertheless there is a considerable amount of reliable evidence indicating that ground level weather observations are related in some way to radio propagation.
In the "quasioptical" region below 10 meters ordinary optical mirage eRects frequently occur under suitable conditions of temperature gradient

'" H. Nagaoka, Radio Research Japan, Report 2, 49

(1932).

26~ E. Qua*ck, E. N. T. 8, 46 {1931).

'""4

G. A.

XV.
M.

Pickard, Skellett,

Proc. I. R. E. 19, 1 f66 (1931}.
Phys. Rev. 37, 1668 (1931).

""A. M. Skellett, Proc. I. R. E. 20, 1933 (1932).

~66 T. Minohara and Y. Ito, Radio Research Japan,

'" Report 3, 115 (1933}.
J. P. Schafer and %. M. Goodall, Proc. I. R. E. 20,

1941 (1932).

~68 F. A. Lindemann, Nature 118, 195 (1926).

"'2"

H.
J.

B. Maris, Proc. I.
Mascart, Comptes

R. E.
rendus

16, 177 {1928). 198, 544 (1934).

'"""~3VPE...MMOa.. lzMHeuirl,llbmuNarnta,,turePArmo1c.3.2GN, ea1ot3p.7hAy(sc1.a9d3.U3Sn)i.coin.,

19, 34 (1933). Trans. , 13th

Meeting (1932), p. 124

and moisture gradient. As mentioned in Section

C, distant transmitting stations, one or two

degrees below our optical horizon, are occasion-

ally received in Cambridge with phenomenal

intensity. The transmission path is presumably

confined to the troposphere, and the slight

amount of atmospheric refraction involved does

not depend upon the existence of ionization.

In the regions above 10 meters, however,

there is no general agreement upon the explana-

tion of the effects which are observed. Bureau"4 suggests that "the ionized layers of the upper

atmosphere play the principal role; but in
certain conditions. . . a slight modihcation can

decide between two possible and different paths

along these layers. This modification may be the

act of the phenomena of the troposphere which

" would then become the arbiters of the propaga-
tion and would decide the fate of the wave.

Though this viewpoint might explain changes of

signal intensity sometimes noted at sea after

crossing air mass boundaries, it would scarcely

account for the numerous reports of actual altera-
tions in E layer height and electron density
accompanying barometric changes. Ranzi'" 6nds
great increases in E ionization after sunset

when place

barometrical of observation

depressions or north of

oitc.cuRratcaltiRet'he'

agrees with Ranzi's observations and maintains
8 that a large rein cloud may aRect the layer.
Martyn"' reports a close correlation between E

layer ionization density and ground-level baro-

metric pressure 12 to 36 hours later. ColwelP78

claims an accuracy of 85 to 90 percent in weather

pstraetdioicntiso. nsFucbhass'ed'

on reception of broadcasting 6nds that variations in signal

strength depend on atmospheric pressure along

the transmission path.

Thunderstorms seem to produce eRects which

are even more remarkable. In Fig. 13 we have

represented reflection conditions before, during,

and after local thunderstorms which occurred in

"F Cambridge at the point of transmission and
reception of an 86 meter signal. The layer

'" R. Bureau, Comptes rendus 188, 455 {1929}.

~~' I. Ranzi, Nature 130, 545 (1932}.
'" J. A. RatcliEe, Proc. Roy. Soc. 45, 399 (1933). "' D. F. Martyn, Nature 133, 294 (1934}.
"R. C. Colwell, Nature 130, 627 {1932);R. C. Colwell

"' and

A.
J.

W. Friend, Phys. Rev. 50, 632 (1936). Fuchs, Funkmagazin 2, 1021 (1929).

HARRY ROKE M I M NO

height" curve is an average obtained from the
records of 22 thunderstorms. Ten additional
storms took place during the test period but
these have been omitted from the average on
account of insufFicient data, or because the
storm occurred at a time when the F layer was
seriously afkcted by the normal diurnal cycle, or by magnetic storms. The apparent increase
in the "height" of the F layer, at the outbreak
of the storm, might be attributed to decreased
group velocity in the underlying regions which
the signal must traverse.
8 Since vertical incidence layer reflections are
less common than F layer reflections at 86 meters, the curve of E layer height represents
an average obtained from the three storms
which happened to occur while adequate E layer
reflections were present. In complete agreement
with the F layer curve, our E layer observations
indicate a maximum echo delay at the time of the outbreak of the storm. The curves of Fig. 13 also indicate that the break of the storm de-
creases the probability of occurrence of sporadic
E reflections. However, when such reflections do
occur at this time, an abnormally high intensity
is indicated by an increase in the length of the
train of echoes (i.e., in the number of multiple reflections). Such action might be the result of a
general increase in the ionization of the upper
atmosphere, which frequently extends downward
into the "absorbing" D region, but occasionally
does not appreciably affect the levels below the E layer. C. T. R. Kilson and others'" have sug-
gested that the enormous electric 6elds of
thunder clouds may produce a penetrating radiation which could cause ionization at a
considerable distance from the storm center.
Additional experimental data is needed in order
to decide whether this hypothetical radiation
actually exists and to determine its nature.
Although our few individual observations,
made in New England, consistently indicate an
increase in E layer height at the break of the
storm, it is possible that other effects may be
observed under different meteorological con-
ditions. RatcliR'e"' reports a storm which apparently produced a marked temporary lower-
ing of the E layer.
"' R. A. Watson-Watt, Nature 132, 13 (1933).
~"' J. A. Ratclie'e, Science 80, 86 (1934).

Q. LOCAL IONOSPHERIC CLOUDS
In Section K reference was made to the fact
that the ion density in the E region does not
decrease steadily and smoothly after sunset as a
simple recombination hypothesis might suggest. Numerous observers have noted high night ionization'" and the very sudden appearance of
strong nocturnal E layer reflections'~ which do
'" not appear to be associated with local storms or
general magnetic disturbances.
Frequently the ion densities suddenly attained
'" on such occasions exceed the greatest densities
reached during the day. Our detailed analysis of the prevalence of
8 86 meter layer reflections in New England is
presented in Fig. 14. An ordinate of 10 percent
indicates that E reflections were present on onetenth of the occasions observed. It will be noted
that the daytime values are low in winter and high in summer. The depression during midday hours is presumably due to absorption in the underlying D region. (In agreement with this hypothesis it may be noted that F& layer reHections are also less prevalent at noon than in the forenoon and afternoon. ) The probability of
observing 86 meter E layer reflections is a
maximum in midsummer, but there is also a subsidiary maximum in midwinter. This winter
maximum is contributed largely by nocturnal ionization. There is a small hump in the daily prevalence curve shortly before sunrise. This occurs both in summer and in winter, and has
been observed during two successive years.
No adequate explanation of this presunrise effect has been brought to our attention,
The sudden and frequent appearance of very
strong E layer reflections of short duration
might be due to a violent momentary increase in the activity of some corpuscular ionizing agent capable of acting over a wide area on the dark
side of the earth. It might equally well be due to
a dense moving cloud of ions, with relatively sharp boundaries of limited extent, which happens to drift over the local area where the
'"T. R. Gilliland, Nat. Bur. Stand. J. Research 11, '" i41 (1933).
H. R. Mimno and P. H. Wang, Phys. Rev. 45, 291
"' (1~93' 4H).. Mogel, Telefunken Zeits. 14, 21 (1933}.
E. V. Appleton, R. Naismith, and G. Builder, Nature
132, 340 (1933).

Flu. 13.

PH YSI CS OF THE IONOSPHERE

F LAYER HEIGI lT
XII.ONI ,'f 888

NUMKR Ol' MULTIPLE
R EFLEP ONS
a~--

CLOea

50
PRF' IALEN( E OF
E II EFLEC"ON
40
I'EAOEN'r

30

o' "~'

20

0 -6
SO

~2

+4

TIME IN HOURS

~6

0-6

AFTER BEGINNING

-2
OF THUNDERSTORM

IONIZED LAYER ACTIVITY AND THUNDERSTORM'

39
+4

10
Fro. 14.
K REFLEXTION
2C:—

MARCH
P

JULY TINK

o IlAT 4IINC ~ JIFFY

0

oo

f
0

NOVQIKR

oI

o~o 0

o

IC

0~

0

- 0 0 0 NOK ONO ~ i IAN o ~, 8l Nfl

OO

2

4

6

6

IO

™ l2

l4

TINK HOURS

l6

N

20

experiments are being performed. Our own
preliminary experiments, conducted simultane-
ously at Cambridge and at Worcester, tend to confirm the second hypothesis. There appear to be measurable time difkrences, of the order of one minute, consistently indicating a progressive motion of such a cloud over the 40 mile inter-
vening distance.
If these preliminary indications could be confirmed they might prove to be of considerable importance in connection with protection of life

and property. The original aircraft radio beacons
" were subject to serious errors caused by the
reHection of an undesired "sky wave. This condition was greatly improved by a modification in the design of the antenna system, which decreased the energy radiated upward. Nevertheless, even the improved antenna system must radiate upward to a moderate extent, and it is at least conceivable that sudden reHections from small dense local clouds of ions might occasionally cause fatal local deviations of short

40

HARRY ROKE M I M NO

duration at a particular spot on the beacon
" range. Aircraft pilots mention such "dead
spots. The reports of recent disastrous crashes indicate that experienced pilots sometimes lose their way, without apparent reason, at times when their beacon is producing a satisfactory signal at distant ground stations, and is properly guiding aircraft over other portions of the same route.
Unfortunately this experiment has been completely interrupted for a period of nearly three years by a remarkable type of quasilegal contro-
'" versy unexpectedly initiated by the Federal
Communications Commission. In order to terminate a long protracted argument involving merely the precise meaning of the phraseology of the Radio Act, it has now become necessary for us to seek a special Act of Congress specifically authorizing use of the automatic devices appropriate to our research No scientific objection to their use has ever been offered, since it is admitted that our methods conform to reasonable engineering practice in all respects.
Further indications of an unfortunate lack of governmental cooperation in fundamental research appear in the frequency assignments offered to experimental stations. In the range of wave-lengths extending from j.5 to 30,000 meters, less than one-half of one percent of the spectrum is available for general experimental use. Such assignments are thoroughly disproportionate and inadequate. The present practice allows legitimate scientific research to be crowded out by adjacent services having greater numerical strength. Such a policy is obviously completely short-sighted and against public interest, butit cannot be adequately combated by individual physicists.
R. ScATTERING oF RADIo WAvEs
In Section C we have referred to a "silent zone" or "skip region" which surrounds transmitting stations operating on a sufficiently high frequency. Points in this area lie beyond the range of the ground wave but are not distant enough to receive the usual sky wave from the
E or Ji regions. This is not always a zone of
s6H R M;mno Sc,ence 8g 54 (193

complete silence, however, for steady signals of

moderate intensity sometimes appear at points

which should normally be inaccessible.

An experienced radio telegraph operator often

realizes that these signals possess some unusual

character. The dots and dashes have a hollow

ringing quality reminiscent of footsteps echoinp

from a stone staircase. By transmitting ground

wave signals simultaneously on a longer wave-

length, Taylor and Young'" were able to show

that these freak echoes are abnormally delayed.

On arrival at a point only 420 km from the

transmitter the echoes had various time lags

corresponding to 2000 to 10,000 km of path

length. At First they ascribed these echoes to

signals scattered backward into the "silent zone"

by distant mountain ranges. On the basis of

directional observations they later decided'"'

that the distant scattering object might be the

'" sea. Eckersley'"

suggests that the ionosphere

has a cloudlike structure which might produce a

complex scattering effect. He also suggests'"

that scattering may be due to multiple splitting.

Howe'" has collected a number of observations

on this subject, including Mogel's opinion that

directive beams produce scattering only at the

top of the ray path. Our own observations""

definitely indicate the presence of nonuniform,

cloud-like ion formations, and tend to prove that

the scattering centers are in the ionosphere and

not on the earth. The ion formations observed

at great distances from our transmitter seem

capable of producing the type of echo observed
by Taylor and Young at times when the E and

Ii layers are readily penetrable.

S. INTERAcTIQN oF RADIo WAvEs
In the usual mathematical treatment of radio propagation it is tacitly assumed that the ionosphere is a linear transmitting medium, and that the behavior of a designated wave is therefore unaffected by other waves which happen to be passing through the same region of space at the same time. Considerable interest

s 7 T. L. Eckersley, Nature 122, 245 (1928).

'"mssT
'"' T.

L L.

Eckersley, Eckersley,

J. Inst. Elec. Eng. 67, 992 {1929).
Electrician 102, 468 (1929).

T. L. Eckersley, Wireless Engineer 6, 255 (1929).

~"' G W. O. Home, Wireless Engineer 8, 579 (1931).

PH YSI CS OF TH E I 0NOSPH ERE

FIG. 15. Reflection conditions during solar edipse of 1936.

was therefore aroused in Europe in 1933 when it was found that the powerful broadcasting station at Luxembourg was producing "cross modulation" which could be heard faintly on the waves emitted by other stations. %hen first reported by Tellegen'" and various broadcast listeners'"' —"-" some skepticism was shown, as it was obvious that a spurious "cross modulation"
might readily take place in the broadcast receiving set.
The reality of the effect was carefully con-
firmed by Van der Pol'" '~ and it was found
that waves passing directly over the powerful transmitter as indicated in Fig. 4 experienced a maximum distortion.
A possible explanation has been proposed by Bailey and Martyn29 3 0 who find that a station of 200 kw operating on a wave-length of 1190 meters can produce an appreciable change in the mean velocity of agitation of the electrons in the ionosphere at points nearly over the transmitter.

B 292
8 293

D. H. Tellegen, D. H. Tellegen,

Nature 131, 840 (1933). World Radio 1V, 165 (1933).

294 Q 295 U

D. R.

SH..IT. eDBisecguesns,ioWn, orWldireRleasdsioW1o8rl,d35335,(1296334()1.934).

B 29B Van der Pol, Science VQ, 11 (1934).

B 29'I
298 V

Van A.

der Pol, U. Bailey and

R. D.

SF..I.M, Laortnydno,n

Congress (1934). Nature 133, 218

(1934).

299 V A. Bailey, Nature 133, 869 {1934).

$00 Q A. Bailey, Phil. Nag; 18, 369 (1934).

This will, in turn, produce a change in the frequency of collision of the electrons with molecules and hence in the absorbing power of that portion of the ionosphere. The absorbing power therefore varies in accordance with the modulation frequency of the station, and so the modulation will be impressed on any other carrier wave which may traverse the region. A quantitative treatment has been presented.
Although there is probably no relation between the two effects, it may be interesting to refer to an experiment performed by Ferrier and Donder'" who were investigating telephony over an ultraviolet and an infrared beam of light. They state that an absorber placed near the transmitter where the rays are concentrated,
has a different effect from the same absorber placed in the weaker radiation field near the
receiver.
T. EcLIPsE OBsERvATIQNs
By means of observations made at the time of a total solar eclipse it is possible to obtain additional experimental evidence in regard to
the nature of the incident solar radiation and the stratification of the ionosphere.
'" R. Ferrier and Th. de Donder, R. G. E. 2'7, 125D
(1930); 2"I, 133D (1930).

HARRY ROWE M I M NO

&!4Iii„

,
rslr wsa'

.''~'"

FiG. 16. ReHection conditions during solar eclipse of 1932.

The space model shown in Fig. 15 illustrates the general nature of the eclipse eHect. In this space model, which will soon be described in
greater detail elsewhere, the X coordinate is the hour of the day (from midnight to noon), the F
coordinate is transmitter frequency (from 2500 to 8500 kc), and the vertical Z coordinate is layer height. The sharp ridge, which occupies the central portion of the model, is the boundary
which separates the Fi reflection from the F2
reflection. The hatched area represents penetra-
tion. Ke may pass from one type of reHection to
the other by varying the frequency or the time of observation. On a normal morning this
boundary ridge would rise steadily. The large
V shaped dip results from the passage of the moon's shadow across the ionosphere. This
depression centers at optical totality, within the
limits of accuracy of the observation.
Fig. 15 has been prepared by Mr. J. A.
Pierce and is based upon extensive data obtained by our group during the June 19, 1936, Russian eclipse by the simultaneous use of the variable frequency and constant frequency methods of observation referred to in Section H. The measurements were made at Ak-8ulak in western Turkestan. Observations of previous

'" "' eclipses'"

though individually less com-

plete, are in good agreement with Fig. 15 when

fitted together. Fig. 16 is a composite space

model of the 1932 eclipse obtained by combining

our own observations with the data reported by

other groups. As this eclipse occurred in New

England in the late afternoon, the time scale

runs from noon to midnight, the frequency

scale from 500 kc to 6500 kc, and the general

trend of the I'&F2 boundary ridge is downward.
I The smaller ridge below the main &F2 boundary
represents the transition from E to I'& reflections.

It was omitted from Fig. 15 as the field equip-

ment taken to Russia was primarily designed to
I investigate the region. The progress of the

sunspot cycle, rather than the geographical

difference, is chiefly responsible for the fact that

'~ K. Maeda, Radio Research Japan, Report 4, 89

'" (1934).
T. Minohara and Y. Ito, Radio Research Japan,

Re'p~orEt .

4, V.

109 (1934).
Appleton and

S. Chapman,

U. R. S. I., London

Co'ng'Sre.ss,S.SeKpitrebmy,berL.(1V9.34B)e.rkner, T. R. Gilliland, and

K.~A" .J.NTor.toHn,endPerrosocn.

I. R.
and

E. 22, 247 (1934).
D. C. Rose, Can.

J. Research

8, '29' J(.19P3. 3S)c. hafer and W. M. Goodall, Science 76, 444

(1932).

3 8 G. W. Kenrick and G. W. Pickard, Proc. I. R. E.

21,' 5' 4(G6 e(n1e9ra3)3)S.urvey), Nature 130, 385 (1932),

PHYSICS OF THE IONOSPHERE

the F~F2 boundary has been shifted to a higher frequency in the 1936 eclipse.
The general simularity of Fig. 15 and Fig. 16 is of considerable interest since it may be shown that hypothetical corpuscular radiation, accompanying the light of the sun, should have produced markedly different behavior in the two cases. The general simplicity of the eclipse eAect seems to indicate that ultraviolet light is responsible for all of the observed phenomena.
The large magnitude of the alteration, and the fact that it is observable with considerable accuracy, indicates that eclipses provide useful quantitative data, available as a numerical test of the accuracy of any ionization and recombination theories which may be presented in order to account for the observed stratification of the ionosphere. These quantitative observations
'" should be continued and it is possible that partial
eclipses will also be of interest.

U. CONCLUSION
In this relatively new branch of physics it is possible to perf'orm a variety of experiments, some of which yield precise numerical data. The
physical system under examination is exceedingly complex and must be studied by statistical methods. Classical magneto-ionic theory offers satisfactory quantitative interpretations for many individual experimental facts, though there are some anomalies which require further attention. No thorough quantitative study of molecular or atomic ionization and recombination processes, designed to explain the observed stratification of the ionosphere, has yet produced any generally accepted comprehensive theory. However, in view of the detailed experimental knowledge already available this major theoretical problem
is worthy of serious attention at the present
time.