Some Equivalence Theorems of Electromagnetics and Their

Application to Radiation Problems

By S. A. SCHELKUNOFF
After a review of the general aspects of the classical electromagnetic
theory several .. equivalence" theorems are established and illustrated with
a number of examples from the diffraction theory. Then follows a discus-
sion of possible applications of these theorems to radiation problems. The
latter part of the paper is devoted to the calculation of the power radiated
from an open end of a coaxial pair.

T HEcircuit
usual methods of calculating the power radiated by an electric
depend upon a determination of the electromagnetic field
from the electric current distribution in the circuit. The best known
of these methods consists in integrating the Poynting vector over the
surface of an infinite sphere surrounding the circuit. This method has
been used exclusively until recent years; to facilitate its application,
John R. Carson obtained a compact general formula for the radiated
power. 1 Another method 2 consists in ca.Jculating the work done
against the forces of the field in supporting a given current distribution
in the circuit. Theoretically either of the two methods is sufficient
for solving any radiation problem. Practically, aside from inherent
difficulties involved in the calculation of the electric current distri-
bution in the first place, the preliminary integration for determining
the field components E and H may be rather complex. Thus in
obtaining the power radiated by a semi-infinite pair of perfectly
conducting coaxial cylinders this preliminary integration has to be
extended over the infinite surfaces of the two conductors. And yet by
the Maxwell-Poynting theory, no energy can flow through the walls of
the outer cylinders since the electric intensity E and hence the Poyn ting
vector vanish there. Any energy which is radiated away must pass
through the open end and it is natural to expect that there must be a
method for calculating this energy from the conditions at the open end.
The integration involved in this method would extend only over a
comparatively small area of the open end. It is in search of a method
of this type for calculating the radiated power that I was led some time
ago to certain" equivalence theorems." Subsequently I learned that
1 John R. Carson, .. Electromagnetic Theory and the Foundations of Electric
Circuit Theory," The BtU Sysltm Technical Journal, pp. 1-17, January 1927.
I A. A. Pistolkors, "The Radiation Resistance of Beam Antennas," Proc. I. R. E.,
Vol. 17, No.3 (1929). R. E. Bechmann, "On the Calculation of Radiation Re-
si8tance of Antennas and Antenna Combinations," Pro,. I. R. E., Vol. 19, p. 1471
(1931).
92
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 93

one of these theorems was discovered long ago, first by A. E. H. Love a

and then by H. M. MacDonald • and proved by the latter 0 for the case
of non-dissipative media in 1911. Another proof of this theorem,
believed to be helpful from the physical point of view and extended so
as to include the dissipative media, is given in this paper. After a
brief review of some fundamental concepts we shall prove these
equivalence theorems, discuss their significance, and solve one or two
simple examples for illustrative purposes.
The physical sources of electromagnetic fields are electric and mag-
netic charges in motion, that is electric and magnetic currents. The
radio engineer has never been interested in shaking magnets for the
purpose of radiating energy and has settled into a habit of ignoring
magnetic currents altogether as if they were non-existent. It is true
that there are no magnetic conductors and no magnetic conduction
currents in the same sense as there are electric conductors and electric
conduction currents but magnetic convection currents are just as real
as electric convection currents, although the former exist only in
doublets of oppositely directed currents since magnetic charges
themselves are observable only in doublets. And, of course, the
magnetic displacement current, defined as the time-rate of change of
the magnetic flux, is exactly on the same footing as the electric displace-
ment current defined by Maxwell as the time-rate of change of the
electric displacement. We shall find it convenient, at least for
analytical purposes, to employ the concept of magnetic current on a
par with the concept of electric current.
The two fundamental electromagnetic laws can now be stated in a
symmetric form. Ampere's law as amended by Maxwell is: An
electric current is surrounded by a magnetic field of force,' the induced
magnetomotive force in a closed curve is equal to the electric current passing
through any surface bounded by the curve. In its original form, the
"electric current" meant only the conduction current so that the law
was applicable only to closed conduction currents. Maxwell's amend-
ment consisted in including the displacement currents, thereby making
the law applicable to open conduction currents. The second law is due
to Faraday: A magnetic current is surrounded by an electric field of force;
the induced electromotive force in a closed curve is equal to the negative of
the magnetic current passing through any surface bounded by the curve.
The rule for algebraic signs is as follows: choose some direction of the
closed curve as positive and have an observer placed in such a way that
a A. E. H. Love, "The Integration or the Equations or Propagation or Electric
Waves," Phil. Trans. A, Vol. 197, pp. 1-45 (1901).
4 H. M. MacDonald, "Electric Waves," p. 16 (1902).
I H. M. MacDonald, "The Integration or the Equations or Propagation of Electric
Waves," Proc. London Mathematical Society, Series E, Vol. 10, pp. 91-95 (1911).
 94 BELL SYSTEM TECHNICAL JOURNAL

this direction appears to him counterclockwise; then the positive

direction of either the electric or the magnetic current is chosen toward
the observer. If the currents are flowing toward the reader, the
directions of the E.M.F. and the M.M.F. are as indicated in Fig. 1.
u v

Fig. I-The relative directions of the E.M.F. and the M.M.F. induced respectively
by the electric current I and the magnetic current K are indicated by the arrows.
Both I and K are directed toward the reader.

In the well-known way these two physical laws lead to a pair of

partial differential equations
curlE = - M, curiH = J, (1)
where J and M are respectively the total electric current density and
the total magnetic current density. The electric density is composed of
several parts; namely: the conduction current density, the displace-
ment current density and the applied current density. The first of
these components is, in many substances, proportional to the electric
intensity E; the second is proportional to dE/dt: and the third is due to
forces other than those of the field, mechanical or chemical, for
instance. Similarly the magnetic current density is the sum of the
magnetic displacement density proportional to dH/at and the impressed
magnetic current density. Thus, we write
aH aE
curIE = - Mo - J.i.Tt, curlH = Jo+ gE + Eai' (2)

where J o and Mo are the densities of the impressed currents and the
constants of proportionality g, E and J.i. are respectively the con-
ductivity, the dielectric constant and the permeability.'
The functions J o and Mo are supposed to be known functions of
coordinates and of time, representing the distribution of the physical
• A consistent practical system of units is used in this paper. Thus the E.M.F.
is measured in volts, the electric current in amperes, E in volts per centimeter, H
in amperes per centimeter, etc. The permeability of vacuum is then 4rtO-1 henries
per centimeter and the dielectric constant (1/3611')10-11 farads per centimeter.
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 95

sources in the space-time. If they are zero everywhere and at all

times, the only physically significant solution of (2) must be E = H
= 0 throughout the entire space a~d at all times. 7 If there are other
solutions of (2), they are extraneous and some rule must be found for
excluding them. Such extraneous solutions often find their way into
mathematical equations because it is usually impossible to express all
physical conditions by an equation or a system of equations. Naturally
these remarks do not apply to a limited region of space or a finite
interval of time. In fact, in many physical problems these "extrane-
ous in the large" solutions of (2) can be advantageously used for
expressing the general character of electromagnetic phenomena in a
limited region and then obtaining, with the aid of the boundary and the
initial conditions, the complete answer. But the philosophy of
causality demands the dictum "no sources, no field" when considering
the whole space-time. It may seem unnecessary to dwell at length on
such obvious matters but they happen to be essential in the subsequent
discussion if. the arguments are to be taken as positive proofs rather
than as plausible justifications.
Equations (2) are linear and the principle of superposition is
applicable. This is in accordance with physical intuition which tells us
that we can subdivide the impressed currents into elementary cells of
volume dfJ, calculate the field due to a typical element, and obtain the
total field by integration. For the typical element (2) becomes
aH iJE
curl E = - J.I. at ' curl H = gE + Eat, (3)

everywhere except in the infinitely small volume occupied by the

element. The product of the current density and the volume of the
element is called the moment of the element.
At times the impressed currents are confined to sheets so thin that
their thickness can be disregarded without introducing a serious error
in the result. This leads to a hypothetical infinitely thin current sheet.
We pass from a real current sheet to an ideal one by assuming that the
thickness of the former decreases and the current density increases in
such a way that their product remains constant. This product is
called the linear density of the sheet and it represents the current per
unit length perpendicular to the lines of flow. The moment of a
current element is now the product of the density of the sheet and the
area of the element. Finally if the impressed current is confined to a
7 We aB8ume that all the electric and magnetic charges were originally in the
neutral state, in which case their separation could be effected only through their
motion. The argument could be extended so as to include purely static fields that
may constitute an integral part of the universe but it is of no particular interest to us.
96 BELL SYSTEM TECHNICAL JOURNAL

very thin filament, the moment is the product of the current and the
length of the element.
It is the moment of the current element that determines its electro-
magnetic field. If the medium is non-dissipative, the actual expres-
sions for the field components are obtained in terms of an auxiliary
function caIIed by Lorentz the retarded magnetic vector potential. For
an electric current element of moment pet) this vector potential at any
point P is parallel to the current density and is a function of the
distance r from the element to P

(4)

The quantity c has the dimensions of a velocity and it appears that the
action of the source travels outward with this velocity. But there is
another solution of (3)

(5)

One might wonder if this solution appertains in any way to the source;
that is not the case, however. If the moment p(t) is identically zero
prior to some instant t = to, the field which can legitimately be attrib-
uted to the action of this source is also identically zero for any instant
t < to. But (5) implies a non-vanishing field at distant points; it is as
if the effect appeared before the cause. Any other solution is a
combination of (4) and (5) and has to be rejected on the same grounds.
In terms of this auxiliary vector potential the field components can
be expressed as follows

H = curiA,
aE 1
-at = -curiH E =
1t
L", curl H dt.
~ (6)
E '

If the moment is harmonic of frequency j, we regard it as the real

part of pel.". Then the vector potential and the field components are
the real parts of the following expressions

H = curl A, E = -
.
~WJ.l.
A + grad• div A , (7)
~WE

where the phase constant

A is the wave-length, and the time factor el ." is implied.

 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 97

If the medium is dissipative, we have

pe-tlf'
A=- H = curiA, E = - iwpA + grad d~v A. (8)
47rr ' g + UoIE
The quantity u is the intrinsic propagation constant of the medium and
is defined by
u = ..JiwJ.I(g + iwe). (9)

In this case the action of the source at some point is not only delayed
by the time needed for the disturbance to travel the intervening
distance but also exponentially attenuated.
If instead of an electric current element, we are dealing with a
magnetic element, the field components can be expressed in terms of an
auxiliary electric flutor potential.' This vector F is given by

(10)

where the moment P of the element is the product of the magnetic

current density and the volume of the element. The field components
are then given by
E = - curl F, ·)F
(+ ~we
H =-g + grad.div•
F (11)
$WJ.l

In the periodic case the general mathematical solution for the vector
potential of an element is found to be a linear combination of any two
of the following functions
cosh ur sinh ur
(12)
r r

AJI of these except the first become exponentially infinite at an infinite

distance from the source and cannot be taken to represent the vector
potential of a physical source. The last function is finite in any finite
region; conceivably it can represent an electromagnetic field in the
finite region free from physical sources. If the medium is non-
dissipative it is impossible to exclude any of the solutions given by (12)
on the grounds of their behavior at infinity-they all vanish there.
But we may regard the non-dissipative case as the limit of the dissi-
pative one and in this way establish a rule for finding the proper unique
solution.
In the presence of a current sheet, equations (2) are valid on either
side of it but not on it. Let us consider a cross-section of an electric
current sheet, perpendicular to the lines of flow, and a curvilinear
 98 BELL SYSTEM TECHNICAL JOURNAL

rectangle A'B'B"A" with two of its sides parallel to the sheet (Fig. 2).
We assume that the current flows toward the reader and that A'A"
and B'B" are vanishingly small. Since the M.M.F. around this
rectangle is equal to the electric current passing through it and since
this M.M.F. is merely the difference between the M.M.F.'s along the
sides A'B' and A"B", we obtain
HI' - H/' = J, (13)

by simply calculating these quantities per unit length of the rectangle.

The tangential components of the magnetic intensity are regarded as

e"

s·

I
// " ,,:
I
I
I
I
I
I
I
I

Fig. 2-A cross-section of a current sheet perpendicular to the lines of flow. The
positive direction of the current is toward the reader.

positive when directed from A to B. Thus the tangential component

of the magnetic intensity is discontinuous across an electric current
sheet and the amount of the discontinuity is equal to the density of
the sheet.
Similarly across a magnetic current sheet the tangential component
of the electric intensity is discontinuous and the amount of this dis-
continuity is equal to the negative of the magnetic current .density of
the sheet; thus
Et' - E/' = - M,. (14)

In deriving equations (2) it is also necessary to assume that g, Il

and E are continuous throughout the region under consideration.
They have no meaning on the boundary between two different media.
Since the boundary is a geometric surface, it cannot constitute either
an electric or a magnetic current sheet. Hence the components of
E and H tangential to such a boundary are continuous across it.
These boundary conditions provide a link between the fields in the
two media.
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 99

Let us suppose that we have a continuous distribution of sources

on a closed 'surface C (Fig. 3) and that there are no other sources.
We assume that the sources are harmonic of frequency f. The elec-
tromagnetic field ~ produced by these sources can be calculated
directly from this distribution with the aid of the above mentioned
vector potentials. On the other hand, we can reason as follows.

Fig. 3-A cross-section of a closed surface C.

There are no sources either inside or outside of C; hence everywhere

except on C, we have
curl E = - iwp.ll, curl H = (g + iWE)E. (15)

In the region inside of C we take that solution of (15) which is finite

throughout this region and outside of Cwe select the solution vanishing
at infinity. Both solutions will contain constants which can be deter-
mined from conditions (13) and (14) across the surface. The field
~' obtained in this manner is identical with ~ because the difference
~ - ~' is everywhere source-free and thus must vanish.
Let us now reverse the process and, instead of starting with the
known distribution of sources on C, suppose that we know the field
and wish to find its sources. Let the known field ~ be source free
everywhere except on C. In order to determine these sources S we
merely calculate the discontinuities in the tangential components of
E and H across C. We can utilize this result to establish the major
Equivalence Principle. For the outside portion of ~ we can choose
the outside portion of the field ~' produced by a given system of sources
S' situated inside C and for the inside part of ~ we take any field which
is source-free there. The latter may be, for instance, the inside
portion of the field ~" produced by some sources SIt situated outside C.
Thus we arrive at the following Equivalence Principle discovered by
Love and Macdonald 8; a distribution of electric and magnetic currents
on a given surface C can be found such that outside C it produces the
8 See references 3 and 4 and also H. M. Macdonald, II Electromagnetism" (1934).
100 BELL SYSTEM TECHNICAL JOURNAL

same field as that produced by given sources inside C; and also the
field inside C is the same as that produced by given sources outside C.
One of these systems of sources can be identically equal to zero.
The actual calculations are made as follows. From the discon-
tinuities in the tangential components of E and H, we obtain J and
M by (13) and (14). From these currents we find the two vector
potentials
A = -.!..
411"
II J(r, y', Z/) e-~r dS
r '

-.!.. II
(C)
(16)
F = M(x, 1, Z/) e-~' dS
411" r '
(C)

where r = -.I(x - x')' + (y - y')2 + (8 - s')2 is the distance between

a point P(x, y, z) somewhere in space and a point P'(x, y', z') on C.
From these potentials we calculate the electric intensity and the
magnetic intensity by

E -_ - swp.
. A + grad. div A - curI F ,
SWE
(17)
grad div F . F
H = curI A + .
Swp.
- ~WE •

The proof of the Equivalence Principle can be modified so as to

throw some additional light on it. Let us suppose that given sources
S' are inside the closed surface C and let us make our new synthetic
field by obliterating the old field outside C and leaving everything as
it was inside C. The new field has the same sources S' and besides it
is discontinuous across C. These discontinuities are the additional
sources S whose densities are calculable from (13) and (14). Since
the new field is identically zero outside C, the field produced by Sis
such as to cancel the field produced by S' outside C. Thus the system
of sources S acts as a perfect absorber for the electromagnetic wave
produced by S'. Reversing the directions of the current distributions
on C, we conclude that the system of sources - S produces outside
C exactly the same field as S'.
The Equivalence Principle is closely related to another theorem
which we may call the Induction Theorem. Let us suppose that a
closed surface C subdivides the entire space into two homogeneous
media and that a system of sources S is given in one of those regions
(Fig. 4). Let E, H be the field due to these sources on the assumption
that the medium inside C is the same as that outside. The true field
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 101

outside C must vanish at infinity but it need not be the same as E, H;

let it be E + E', H + H'. The field E ' , H' must be source-free out-
side C. Inside C the field must be source-free; we shall designate it
by E", H". The field E', H' is called the reflected field and E", H" the
refracted field. The boundary conditions are such that the components
of the electric and the magnetic intensities tangential to C must be
continuous. Thus over the surface C, we have

E, + Be' = B,", ii, + H,' = H,". (18)

The bar over the letters is used to designate the values of the corre-
sponding quantities on C. From (18) we obtain

B," - E,' = E" H," - He' = H,. (19)

S(E,H) *
*
*

Fig. 4-The closed surface C is the boundary between two homogeneous refions in
space. (E, H) designates the field produced by some system of sources S; (E I H') is
the field reflected by the body C; and (E", H") is the field in the body.

Hence the reflected and the refracted fields together constitute an electro-
magnetic field in the entire space; this field is source-free everywhere
except on C and the distribution of sources on C is calculable from the
given sources S. This Induction Theorem is a generalization of the
well-known theorem used in calculating the response of an electric
circuit to an impressed field. Since the wires constituting the circuit
are very thin, only the tangential components of E in the direction
of the wires need be considered.
It may be noted that if the medium inside C is identical with that
outside C, the .. reflected field" must be absent and the .. refracted
field" must be identical with the field E, H due to the sources S.
Thus the Induction Theorem leads to the Equivalence Principle.
The Equivalence Principle is evidently an extension of Kirchhoff's
theorem. The latter deals with a single wave function instead of two
vectors. Kirchhoff derived a formula for computing the wave function
in the source-free region from its values and the values of its normal
derivative over a closed surface separating the source-free region from
102 BELL SYSTEM TECHNICAL JOURNAL

the region containing the sources of the wave functions. In the

Theory of Sound the wave function represents the excess pressure or
the velocity potential and Kirchhoff's theorem is valuable in the
analysis of diffraction phenomena. Kirchhoff's theorem is also used
in dealing with optical diffraction. We may also remark that Kirch-
hoff's formula is a mathematical expression of a principle governing
compressional wave motion. This principle was first formulated by
Huygens in the following form: each particle in any wave front acts
as a new source of disturbance, sending out secondary waves, and these
secondary waves combine to form the new wave front. 9
Let us now examine one of the familiar diffraction problems in the
light of the Equivalence Principle. Consider a source S and a per-
fectly absorbing screen (Fig. Sa). Such a screen will be defined in the
usual manner: the impressed wave enters it without. reflection but
does not pass through it. If the screen is infinitely thin, this definition
implies the existence of electric and magnetic currents in the screen
whose densities are given by the postulated discontinuity in the field.
In reality the black bodies" absorb not by virtue of the coexistence
II

of electric and magnetic currents but by virtue of electric currents

alone with the aid of reflections taking place between atomic layers.
The true mechanism of absorption is complex and requires more than
a mere surface. In diffraction studies it has become a habit with us
to ignore the precise nature of absorption and confine outselves to its
implications; but it is just as well to know the nature of the ideal
mechanism which we are substituting for the true mechanism.
We can apply the Equivalence Principle to the present problem in
two ways. We can choose as our surface C a surface (1234) just on
the other side of the screen. The part (23) contributes nothing; the
equivalent distribution of sources S' over the parts (12) and (34) gives
us a complete field to the right of the screen. On the other hand if
S" is the field due to the electric and magnetic currents in the screen
induced by S, the total field is S + S", The choice of the II surface
cor that would yield this result is shown in Fig. Sb although the con-
clusion is obvious without recourse to the Equivalence Principle.
Since to the right of C in Fig. Sa the two alternative fields must be
the same, we have

S' = S + S" and S' - S" = S. (20)

Incidentally the last equation is the expression for the Equivalence

Principle as applied to S in the absence of the screen since - S" is
9 A. E. Caswell, II An Outline of Physics," p. 544 (1929).
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 103

the contribution of that portion of the equivalent layer which was

removed by the screen.

s*

3
---------------------4
2

Fig. 5a-A source S in front of a screen the cross-section of which is

6hown in heavy lincs.

:::::.:.:::_:::::.-:::.::.:.::. *
5
11 10 '~-"9 8
-_-::::::.:.-:..~ ~:::.::-_-_-:::::_-_-_-_--:..-_--:..-_-_-.::-.::-:::_-_-:::::~ 6
I 2 , __ " 3 4

Fig. 5b-A source S in front of a screen the cross-section of which is

shown in heavy lines.

The case of a hole in a perfectly absorbing screen (Fig. 6a and Fig.

6b) can be treated in the same manner and the reciprocity existing
between this and the preceding case is quite evident. In terms of
the sources previously defined the field to the right of the screen is
- S"; by (20) this is the same as S - S'.
If the screen is a perfect conductor, the problem is much more
complex. The screen will support electric currents but not magnetic
currents. The densities of the electric currents are not calculable
directly from the field S but from the condition that the component of
the electric intensity tangential to the screen vanishes. The problem
104 BELL SYSTEM TECHNICAL JOURNAL

is very difficult and its solution has been found in only a few special
cases. It is true that once we know the electric currents in the screen,
we can detennine the field on both sides of the screen; but there is no
simple way of calculating these currents exactly. Frequently it is
assumed that, in so far as the side opposite to the source is concerned,
a perfectly conducting screen is equivalent to a perfectly absorbing
screen of the same geometric character. This is equivalent to a

s*

2 J 4

Fig. 6a-A source S in front of a screen the cross-section of which is

shown in heavy lines.

---------
J
----_-~ ~~~~ ====~\ 2 s~
r----------------------
4
____________________ _
6

Fig. 6b--A source S in front of a screen the cross-section of which is

shown in heavy lines.

hypothesis that the electric current density of the screen is deter-

mined by the magnetic intensity impressed directly by the source S.
We could take the results obtained from this hypothesis as a first
approximation to the true results. The tangential component of the
electric intensity calculated on the basis of this hypothesis does not
vanish on the screen which means that we have violated the original
hypothesis that the screen is a perfect conductor. If the discrepancy
is not too great we might look for an additional electric current
distribution to reduce this discrepancy.
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 105

There are times, however, when the current distribution in the

"screen" can be determined with a fair accuracy without elaborate
mathematics. It is so, for instance, in the case of a pair of perfectly
conducting coaxial cylinders (Fig. 7a and Fig. 7b) in which the radii

4 3
...
--,;;.::.=.::.=.:~-~-;.;;-;.;;-;.;;-;.;;-;.;;-~-;.;;-;.;;-;.;;-;.;;-;.;;-;.;;-;.;;-;.;;-;.;;-;;.;-;.;-;.;-;.;-;;;.;;;-..;-----~-------..;.-;;.-.;.---;;.-;;.-;;.-;;;.;;;.-------~- -'j

-----------------------------
I
----- -----
Fig. 7a-An axial cross-section of a coaxial pair.

4
_..... _---

Fig. 7b-An axial cross-section of a coaxial pair.

are small by comparison with the wave-length. We shall assume that

the coaxial pair is semi-infinite. Trusting his common sense, the
engineer assume!> that inside this structure the magnetic lines are
circles coaxial with the cylinders. The electric lines are the radii and
the electric current in the cylinders as well as the transverse voltage
between the cylinders vary along the length in the same way as in a
transmission line with uniformly distributed series inductance and
shunt capacity. A careful analysis by John R. Carson indicates that
this simple picture is justifiable if the cross-section of the coaxial pair
is small by comparison with the wave-length. 1o While a whole series
of electric waves can exist in such a structure, all of these waves except
the one recognized by the engineer, the principal wave, are attenuated
very rapidly and are significant only very close to the generator and
10 John R. Carson, "The Guided and Radiated Energy in Wire Transmission,"
A. I. E. E. Journal, pp. 908-913, October, 1924.
106 BELL SYSTEM TECHNICAL JOURNAL

very near the open end. The complementary waves are needed only
for logical consistency and to satisfy the boundary conditions.
Thus let us suppose that the field distribution in the coaxial pair
is known to a high degree of accuracy. In order to calculate the field
outside the coaxial pair and hence obtain the radiated power we can
use the Equivalence Principle in two ways. We can fit our surface C
smoothly over the outer cylinder and the open end (Fig. 7a) or, re-
garding this surface as a perfectly elastic rubber sheet, we can press
it through the open end and fit it smoothly over the inner surface of
the outer conductor and the outer surface of the inner conductor
(Fig. 7b). Since by hypothesis the conductors are perfect, the com-
ponents of E tangential to the cylinders vanish; hence in the second
choice of C the equivalent layer consists of only an electric current
sheet. Naturally this current distribution is precisely that which
actually exists in the conductors 50 that this choice of C leads to some-
thing that we knew beforehand, namely: if the actual sources, that is,
if the electric currents in the structure are known exactly or approxi-
mately, the entire field can be calculated exactly or approximately.
The first choice of C is more important. Over the lateral portion
(12, 34) of C the equivalent magnetic current sheet vanishes as in the
preceding case on account of the perfect conductivity of the cylinders.
The magnetic intensity just outside the coaxial pair is also zero except
near the open end where it must be exceedingly small. To see this,
we need only recall that the electric currents in the two cylinders are
equal and opposite and that except in the neighborhood of the open
end the displacement currents are transverse. Thus the equivalent
electric current sheet can be ignored altogether. What is left is the
magnetic current sheet over the surface of the open end; the density
of this sheet is determined by the radial component of the electric
intensity and in the final analysis by the voltage existing between the
ends of the inner and outer conductors. Presently we shall carry
out the actual calculations but just now we shall examine the question
of the accuracy of the results. Of course, the results would be exact
if we knew the equivalent electric and magnetic sheets accurately;
and the above approximations appear to be reasonable. We shall not
be able to find out how good these approximations are but we can
prove that they are just as good as the approximations usually made
in calculating the radiated power from the distribution of electric
currents. The only virtue of the Equivalence Principle is to save a
certain amount of mathematical work and furnish a further insight
into the phenomena of radiation.
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 107

If a progressive wave is advancing from left to right in a semi-

infinite coaxial pair (Fig. 7) and if the generator is at infinity, we can
assume it to be the principal wave. At the open end this wave is
reflected. It is usually assumed that the reflected wave is also the
principal wave but moving in the opposite direction. In other words,
it is assumed that the total field is such that the electric lines are
radial and the magnetic lines are circular. Since the electric lines
are radial, there is no longitudinal displacement current; and since
the conduction current at the open end must be zero, the magnetic
intensity is zero over the entire open end. This is what follows if we
neglect the complementary waves.
These approximate results correspond to the exact results in the
following hypothetical situation. If a hypothetical perfect magnetic
conductor is fitted over the open end of the coaxial pair so that it
closes it entirely, then the reflection is complete and there are no
complementary waves. Perfect magnetic conductors are defined by
analogy with perfect electric conductors-the tangential component
of the magnetic intensity vanishes at the surface of the former just
as the tangential component of the electric intensity vanishes at the
surface of the latter. Magnetic conductors support magnetic current
sheets just as electric conductors support electric current sheets. The
densities of the sheets are given by the discontinuities of the tangential
components of E in the former case and H in the latter.
In the hypothetical case in which the open end is closed with a
perfect magnetic conductor, no energy can flow outside the coaxial
pair. This is because the flow of energy is given by IE X H and either
one or the other factor vanishes over the outer boundary of the struc-
ture. The field outside the coaxial pair must now be identically zero.
Our sources are the electric current in the walls of the coaxial pair and
the magnetic currents in the cap. If one field is designated by Sand
the other by S', then S + S' = 0 and S' = - S. Thus the field
produced by the electric currents in the conductors on a supposition
that principal waves alone exist, is the same as the field produced by a
hypothetical distribution of magnetic currents over the surface of the
open end.
Let us examine another case. It is usual to assume that the electric
current in a thin wire (Fig. 8) in free space is distributed sinusoidally.
Experimental evidence shows that the radiated power calculated on
this assumption is very nearly correct. On the other hand the
sinusoidal distribution of the electric current in the wire corresponds to
a hypothetical case in which a perfect magnetic conductor is introduced
108 BELL SYSTEM TECHNICAL JOURNAL

in the shape of a sphere concentric with the center of the wire and
passing through its ends. Thus we could calculate the radiated power
from an appropriate distribution of magnetic currents over this sphere
but in this case such a procedure would involve more difficult inte-
grations than the usual method.
Before considering the more general case of radiation from a semi-
infinite coaxial pair let us assume that the radii of the two conductors
are nearly equal. We have seen that in applying the Equivalence
Principle we need take into account only the magnetic current sheet
over the open end of the pair. In the present instance this sheet is
merely a circular loop of magnetic current equal to the voltage V
between the ends of the conductors. If we were to treat in the same

---
Fig. 8-A vertical antenna and a cross-section of an imaginary sphere
passing through the ends of the antenna.

manner a condenser made of two parallel circular plates, we should

come to the conclusion that it is also equivalent to a magnetic loop
around its periphery. Thus in both cases the radiated power is the
same. But the power radiated by an electric doublet is known to be
(407r2[2zt)/X2 watts where [is the amplitude of the electric current, I the
length of the doublet and Xthe wave-length. In applying this formula
to a condenser it is better to express it in terms of the voltage V
across the condenser and its area S. The capacity of the condenser is
G = SI(36?r10 11l) farads and I = (dGV = SV!60Xl amperes. Hence
the power radiated by the condenser is ('lr2S2V2)/90X4 watts. This is
also the approximate power radiated by the coaxial pair if we interpret
S as the cross-sectional area of either conductor.
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 109

Let us now calculate the more general expression for the power
radiated from an open end of a coa~al pair. The cylindrical con-
ductors whose cross-sections are shown in Fig. 9 are supposed to extend

Fig. 9-The end view of a coaxial pair of cylindrical conductors.

below the surface of the paper and the z-axis of the coordinate system
is directed toward the reader. The primed letters will refer to points
situated in the opening, the un primed letters being reserved for typical
points in space.
The electric intensity in the coaxial pair varies inversely as the
distance from the axis
P
Ep' E.,. = o.
="p (21)

In accordance with the Equivalence Principle we assume that the field

below the xy-plane is wiped out and the discontinuity in E arising as
the result of the separation is replaced by a magnetic current sheet.
This magnetic current is perpendicular to E; by (14) its density is
P
M.,. = - Ep' = -"
p
(22)

The constant E is related to the voltage V between the ends of the

coaxial conductors; in fact, we have
b V
= P log a'
b
V = [
Ep.dp' P=_·
b
(23)
• a log-
a
110 BELL SYSTEM TECHNICAL JOURNAL

In order to calculate the field at some point A due to the distribution

(20), we must determine the retarded electric vector potential. Since
the integration is vectorial, it is convenient to deal with the cartesian
components of the magnetic current density
_ Psin ",'
M z.I - I (24)
P

The area of the element is p'dp'd",' so that the components of the

rb1
retarded potential are
= J:. b
M",.e-i/JAA'p'dp'dql
411" Ja
rJr
Fz 0 AA'
_ P b b e-i/l AA ' sin ",' I I

- 411" J AA' dpd"" (25)

o
rl>lb e-i/JAA'AAcos rp' dp d", ,
a
_ P W 211"
FI/ - - 411" J /3
I I
= w,fjiE = - = - .
a
0 I c ).

Hence the components in the polar coordinates are

Fop = - Fz sin", + FI/ cos '"
_ P rl> r 2r e-i/l AA ' cos (rp - ",') I ,
- - 411" Ja Jo AA' dpd"" (26)

Fp = O.
The distance AA I is
AA' = ..Jr2 - 2rp' cos {} + p'2, (27)

where r is the distance OA, and {} is the angle between OA and OA I.

Since we are interested in the radiation field alone, we need retain only
those terms in (24) which vary inversely as the distance; the other
terms contribute nothing to the radiated power. Thus we let r
increase indefinitely, obtaining
AA' = r - p' cos t'J, (28)
and then
(29)

If (} and (}' are the angles made by OA and OA' with OZ, we have
cos {} = cos (} cos (}' + sin (} sin (}' cos (rp - ",')
= sin (} cos (rp - rp/). (30)

Since p' is small by comparison with the wave-length >., we can expand
the exponential term in the integrand into a power series and retain
 SOME EQUIVALENCE THEOREMS OF ELECTROMAGNETICS 111

only the first two terms

eiflpl co." == 1 + ifJp' cos f} = 1 + ifJp' sin (J cos (ip - ip'). (31)

We need the second term because the integral of the first vanishes.
Integrating the second term, we obtain

F'I' =
·fJP"- i/Jr •
- ~ ... 4 sin
.".r
(J 1 12r
a
6
'd '
pp
0
cos 2 ( ip-ip')d ip'
(32)
1 e-i/Jr
- "8 i{3(b 2
- a )P -r- sin fJ.
2

The magnetic current is uniform around the axis and there is no

accumulation of magnetic charge anywhere; hence the second term in
the expression for H as given by (11) vanishes. Therefore
P e-i/Jr
H'I' = - gWEfJeb2 - a2) -r- sin 8. (33)

At a great distance from the source the wave tends to become plane so
that in the radiation field the electric intensity is perpendicular to OA
and to H and is given by

E8 = ~~H'I' = 120.".H",. (34)

According to Poynting the radiated power is the real part of the

following integral

W = 2"1 J{"J{''' EBli'l' *r2 sin (J dfJ dip watts, (35)

o o

where H,,* is the complex number conjugate to H",. Substituting from

(31) and (32), we obtain

_ .".a
W - 980
(b 2
-
-}.-2-
a
2
) 2
P
2 . a fJ dfJ J{2r dip
Jr"' sm
o o
=
(b2 _ a2)2
.".4
360 -}.-2- p2 watts. (36)

Introducing from (21) the expression for P and designating by S the

area of the opening, we have

r (
W = 360
}.2
S b
log -
)2 V2 watts. (37)
a
112 BELL SYSTEM TECHNICAL JOURNAL

The effect of radiation on the transmission line can be simulated by a

resistance R,

180( 310g ~)2

X
R =""7 S ohms (38)

shunted across the open end. This is not the resistance seen by the
generator. If V and I are the amplitudes of the voltage and the
electric current at their antinodes and Zo the characteristic impedance
of the coaxial pair, then

V = ZoI = ( 60 log ~) I. (39)

r
Since the end of the coaxial pair is a voltage antinode, the radiated
power may be expressed as

W = 10r ( ~ P watts. (40)

Hence the radiation resistance seen by a generator placed at a current

antinode is
(41)

With this simple illustration, we conclude the present paper.