AUGUST 15, l939 PHYSICAL REVIEW VOLUME 56

On Diffraction and Radiation of Electromagnetic Waves

S. A. ScHELKUNGFF
Bell Telephone I.aborotories,¹m Fork, ¹m York
(Received June 28, 1939)

Inasmuch as it is rarely possible to treat diffraction of equations) depends on the particular choice of the auxiliary
electromagnetic waves exactly, the Kirchhoff formulation vector and in some instances, at least, is obviously un-
of Huygens' Principle has been frequently used in ap- reasonable (Appendix III). The calculations of diffracted
proximate calculations. If the Kirchhoff formula is applied fields and radiation fields, based either on the Equivalence
directly to the field intensities of the incident wave over Principle or on the more general Induction Theorem,
the aperture, the diffracted field is found to be inconsistent depend upon a priori ver@able approximations to the actual
with Maxwell's equations. If, on the other hand, this fields in the neighborhoods of the sources of the diffracted
formula is applied to some auxiliary vector potential from and radiated waves. For this reason we feel that these
which the diffracted field is subsequently deduced by dif- methods are preferable to those based on the Kirchhoff
ferentiation, the result (although consistent with Maxwell's formula.

'HE purpose of this paper is to describe 6eld, that is, the Field which has to be added to
several methods for dealing with diffraction the "impressed" field PE, H' in order to obtain
and radiation of electromagnetic waves, which the actual field in region (1) just within the pipe
are believed to have certain advantages over the when it is terminated; and, Finally, let 8", H"
methods based on the Kirchho6 formula. These be the "transmitted" or the actual Field in region
methods arise out of certain "induction" and (2) just beyond the end of the pipe. Consider the
"equivalence" theorems. At present, it is rarely field E, 0 composed of 8', FI' in region (1) and
possible to solve diffraction and radiation prob- of Z", IX" in region (2). We assert that the field
lems exactly. Either the usual methods or the E, IJ could be produced, in the presence of the
methods to be described in this paper require pipe, by electric and magnetic current sheets
approximations. The point in favor of the latter over the aperture and that the densities of the
methods is our ability to use physical intuition sheets are
and available knowledge, theoretical and experi-
mental, as a guide in making the necessary
approximations. J being the density of the electric current sheet,
The essential feature of the following theorems 3f the density of the magnetic current sheet and
is identiftcation of certain portions of given electro n the unit normal to the sheets pointing into
magnetic fields as the fields of appropriate fictitiotts region (2). Evidently Zo and FIO could be replaced
electric and magnetic current sheets. by their components tangential to the surface
of the sheets.
THE INDUcTIQN THEoREM While a formal proof of this theorem will be
It is not easy to state the Induction Theorem discussed in Appendix I, a few details will be
in its most general form and it is more expedient given here in the belief that they will elucidate
to formulate its variants to Fit special conditions. the meaning of this theorem and the manner in
In one form, the theorem was enunciated in a wh&c»t can be modi6ed to suit different condi-
previouspaper. 'Weshallnowstate itin the form tions. The electric and magnetic intensities of
directly applicable to the problem of radiation. the total Field around the pipe, under actual
from an open end of a perfectly conducting conditions, must be continuous across the aper-
metal pipe. ture; thus over the aperture we have
Let, Z', H' be the Field which would exist over 8'+Z'=E" H'+H'=H";
the surface of the aperture if the pipe were con-
inued to inFInity; let +,
H be the reflecte HI I Ht' Ho (2)
S. A. Schelkunoff, Bell Syst. Tech. J., pp. 92—112,
January, 1936.
3
 ELECTROMAGNETIC WAVES

Maxwell's equations everywhere except over the Induction Theorem, thus affording a veri6cation
aperture and the surface of the pipe. This 6eld of the theorem.
also satis6es the proper boundary conditions over Consider now a system with one degree of
the surface of the pipe. Over the aperture of the freedom. An example of such a system is an
pipe, the 6eld It', H and, in particular, its tan- ordinary transmission line terminated in some
gential components are discontinuous. It follows impedance (Fig. 2). Let V'(x), P(x) be the
from the integral form of Maxwell's equations voltage and the current that would exist in the
(or directly from the laws of Am'pere and Fara- progressive wave moving from left to right if the
day) that discontinuities in the components of line were terminated in its characteristic im-
magnetic and electric intensities, tangential to a pedance instead of the actual impedance. Choose
surface 5, exist if and only if there are corre- some particular point xo on the actual terminated
line. ff V" (x), I"
(x) is the actual voltage and
current to the right of this section and V'(x),
I'(x) is the voltage and the current which must
be added to V'(x), P(x) in order to obtain the
Fir. 1. Pipe enclosing two different dielectric media. actual voltage and current to the left of the
section, then the continuity conditions at x= xo
sponding super6cial electric and magnetic cur- uii e
rent sheets over 5. The density of the electric
current sheet is equal to the discontinuity in the V'(xo)+ V'(xo) = V"(x,),
tangential component of the magnetic intensity I'(xo) + I'(xo) = I"(xo);
ol
and the density of the magnetic current sheet is
equal' to the discontinuity in the tangential V" (xo) —. V'(xo) = V'(xo),
component of the electric intensity. ' I"
(xo) — I'(xo) = I'(xo) (&)
Therefore, if we start with the current sheets
In this case, the Induction Theorem consists in
(1) and wish to determine the field produced by
them in the presence of the pipe, we must regard asserting that the wave V(x), I(x), equal to V',
the 6eld E, H as a possible solution. If we accept
I' on the left of x=xo and to V", I"
on the right
for the moment that any solution of Maxwell's of it, colld be produced by a sero impedance
equations which satisfies given boundary condi- generator in series voith the line and an, infinite
tions and behaves in the proper manner at in- impedance generator (constant current generator)
6nity is unique, then E, H is the 6eld which we in shunt neith the line, the voltage of the series
would calculate from the current sheets (1) no generator and the current througk the shunt
matter what particular method of calculation we generator being, respectively, V'(xo) and P(xo).
are likely to adopt.
Let us consider another situation. Let a EQUIVALENCE THEOREMS
perfectly conducting metal pipe enclose two Broadly speaking, the equivalence theorems
different dielectric media which are separated by differ from the corresponding induction theorems
a plane boundary (Fig. 1) and let a progressive in that the latter specify the current sheets
wave E', II' move from left to right in region (1).
capable of producing both the reflected 6eld 8',
By the Induction Theorem, the wave reflected II' and the transmitted 6eld E", H" while the
from the interface between the two media and
former specify the current sheets capable of pro-
the wave transmitted into region (2) could
ducing only the transmitted 6eld. Thus, if in
be produced by electric and magnetic current
sheets over the interface, whose densities are
given by (1). In this instance, the exact solution tl
v'(x)
can be obtained directly as we11 as from the
' It also follows from the integral form of Maxwell's i'(x) Vo(x)
equations that the discontinuities in the normal compo-
nents of electric and magnetic clrrent densities are deter-
mined by the discontinuity in the tangential 6eld intensities. FIt". 2. Transmission line terminated in an impedance.
310 S. A. SCHELKUNOFF

region (1), we postulate a zero field in place of combine to form the new wave front. I am in-
the reflected field E', H' and two current sheets clined to believe that if with the aid of a hypo-
(over the surface of the aperture) defined by thetical perfectly absorbing screen we could
J"
=n XII", M" = —n XZ", (4)
isolate the disturbance produced by an individual
secondary source, we should find this disturbance
to take care of the discontinuities in the total identical with that produced by the elementary
field B, H comprised of the zero field in region (1) source (4). It is impossible to prove this conten-
and the actual transmitted field E", H" in region tion any more than to prove that the energy flow
(2). It should be noted that in equivalence per unit area in an electromagnetic field is ex-
formula (4) the densities of the current sheets are pressed by EXH. Just as in the latter case, it
not known in advance as is the case in the corre- can be proved only that the total gow of energy
sponding Induction Formula (1). across a closed surface can be obtained by inte-
Another equivalence theorem may be obtained grating the Poynting vector, in the former case
by choosing a closed surface S comprised of the it can be proved only that when we integrate a
surface of the aperture and the outer surface of certain expression over a closed surface we obtain
a horn, postulating a zero field inside the surface the correct field. In either case it is possible to
and Z", IE' outside, and introducing electric and modify the integrand and still obtain the same
magnetic current sheets over S of densities given results when tke integration is performed over the
by (4). We assert, then, that these sheets will closed surface. The choice of a particular inte-
produce the postulated field. This theorem may grand must necessarily be made on other than
be called the "free-space" equivalence theorem mathematical grounds.
in order to emphasize that in carrying out the Let us now consider a perfectly conducting
calculations, the horn must be ignored and the metal tube and a progressive wave moving from
response must be calculated by the "free-space" left to right (Fig. 3). The actual field at some
retarded potentials in contrast with the previous point P to the right of some cross section S~
equivalence theorem according to which the re- could be produced by electric and magnetic
sponse to the current sheets had to be calculated current sheets given by (4), or by (1) since in
subject to the boundary conditions at the surface this case E"=8'
and H"=H'. The action of
of the horn. On the other hand, the current sheets these current sheets must be calculated subject
required by the free-space equivalence theorem to the boundary conditions at the surface of the
extend over the outer surface of the horn as well pipe. This field could also be calculated by the
as over the aperture while in the other theorem free-space equivalence principle by using the
they extend only over the aperture. In a previous current sheets spread over a closed surface S2
paper' the "free-space" equivalence theorem has surrounding point I' and whose densities are
been named the Equivalence Principle' in the still given by (1).
belief that it is a precise expression and a general-
ization of a physical principle enunciated by RADIATION FROM HORNS
Huygens. Preceding theorems can .be used for the ap-
Huygens' principle is commonly stated as proximate calculation of the power radiated by
follows Each element of any wave front acts horns and the radiation patterns of those horns.
as a new source of disturbance, sending out Three methods of approach present themselves:
secondary waves, and these secondary waves
' The Equivalence Principle was originally discovered by ~p
A. E. H. Love (Phil. Trans. Roy. Soc. A19V, 1—45 (1901)j
and subsequently by H. M. MacDonald /Electric Waves
(1902), p. 16).The latter supplied two proofs [MacDonald,
Electric Waves (1902), p. 16; MacDonald, Proc. London
Math. Soc. Series E, 10, 91— 95 (1911)) of the theorem, of
~ P
which only the second is valid. A few years ago, the the- /
l

orem was rediscovered by the present author as one of a

group of theorems treated in a previous paper (reference 1)
and here. FJ6. 3. Perfectly conducting metal tube; progressive wave
4 A. E. Caswell, An Outline moving from left to right.
of Physics (1929), p. 544.
 ELECTROMAGNETIC %AVES 3ii
i. In accordance with the free-space equiva-
lence principle the sources within the horn and
the horn itself' are replaced by electric and mag-
netic current sheets, surrounding the horn, with
densities given by (4); then, the free-space field
of these sources is calculated by means of the
following formulae.

Z= —k&pA+ grad. (div. A) —curl P,

'bo) E'

H= i~a— Frr. 4. Pipe ending in a perfectly conducting plane.

F+ grad. (div. F)+curl A,
'SOAP
It may be pointed out that in theory it is
where the magnetic vector potential A and the possible to calculate the radiation pattern and
electric vector potential F are given by the radiated power from the conduction current
in the horn. In practice, however, no basis has
J"e—'&" M"e '&"
A= dS, I'= —dS. (6) been found for making a reliable a priori as-
(S) 4mr 47rr sumption with regard to this currqnt distribution.
2. The second method of approach consists in
These formulae become much simpler if we are replacing the sources within the horn by electric
interested only in the 6eld at great distances and magnetic current sheets with the same
from the horn. ' densities (4) as in the preceding case but extend-
In practical applications neither nor M" J" ing only over the aperture. These current sheets
are known exactly and suitable approximations are supposed to, act in the presence of the horn
must be made, with available theoretical and as a reflecting surface. Thus, A and F are
experimental knowledge as a guide. The situation given by
is similar to that in which we And ourselves when
calculating the power radiated by antennas and A = J" p, (x, y, s; x', y', s')dS,
(~e3
by transmission lines from the electric current
distribution. The latter is not known exactly but
available evidence points to the fact that it is p= ~" p~(x, y, s;x', y', s')dS,
(~a)
nearly sinusoidal.
If the aperture of the horn is so large that the where P& and il2 are two diadics and the integra-
cut-off frequency of the horn is appreciably below tion is extended over the surface S, of the aper-
the operating frequency, practically all energy ture. The scalar product of the moment of an
reaching the aperture passes into the outer electric current element and diadic Pi is the mag-
medium. In this case, it is reasonable to suppose netic vector poten'tial of the electric current
that the held over the aperture will be sub- element and, similarly, the scalar product of the
stantially the same, except near the edges, as if moment of the magnetic current element and
the horn were in6nitely long. Thus, we can as- diadic P~ is the electric vector potential of the
sume that over the aperture J"
and M" are ap- element.
proximately equal to the known quantities J' In applying (7) we no longer need approximate
and M'. Over the surface of the horn M" is estimates of the 6eld over the outer surface of
known to be very small (it would vanish for the horn; instead we have to approximate P& and
perfectly conducting horns) and is assumed J" P2. In the case of the horn considered above, we
to be small. would say that the effect of the horn, as a reflect-
ing surface, on the radiation held is probably
~
That is, the sources induced in the horn. smaH and that we could replace Pi and P2 by the
6 S. A. Schelkunoff, A General Radictioe Iiornsnla (to be
published). free-space transmission factors. In this case we
 S. A. SCH ELKU NOFF

obtRln thc sRme lesUlt whethcI we follow thc Thus, we have

6rst or the second method of approach.
The second method may, however, be more
valuable than the erst in other situations. For
example, consider a pipe Rnd an infinite perfectly
conducting plane (Fig. 4). The erst method will wliei e i(i aiid @g ai"e tlie same diadics as iil (7}.
give us the same result as if the plane were absent In theory, this theorem is more powerful than
slncc, not, knowing 6 prior J
ln the plane, wc either of the two preceding theorems. It does not
require RIiy RpproxlIIiatlons to the 6eld ovcl th. c
should be forced to ignore it, The second method
will supply us with R better result since we shall surface of integration and it gives both the
deal with electric Rnd magnetic current sheets refIected Rnd thc tIansInlttcd field. II1 practice)
over a perfectly conducting plane. ' This problem however, the major value of this theorem is to
can be solved by the free-space retarded poten- furnish R basis foI RpploxlmRtlng thc field ovcI
tials. The image of the electric current sheet in the aperture. In transmission systems admitting
the plane is negative and wi11 cancel the effect of of only OIie transmission mode, the I rid Uctlon
the sheet itself. The image of the magnetic cur- Theorem leads to a definite solution (see Appendix
rent sheet is positive and will double the CEcct II). We then assume that the known results in
of the sheet. In other words, instead of (6) we the simple case may be taken as a first approxi-
now have mation in more complex cases.

M"e 'P"
P= — d5, A =0.
(5„) 2 et' It 18 only fitting thRt wc shoUM. not pass ln
Since M is detcI rnincd by thc tRngentlal silence the well-known Kirchhoff formula which
components (Z, ", E~") of the electric intensity, has been universally Used for solving di6raction
we can obtain from (5) and (8) problems and recently has been applied to
problems of radiation. ' '0 "This formula is
'&'
1 t' Iye
E ") ip+ cos HdS,
2ir (s i ( r) r ~

V= ( iP+ (Vcos —
(ri, r)—
(~) &

where V is a wave function, 5 any closed surface

surrounding a source-free region, and e is the

Z, = — — (Z," cos q+Z„" sin y)

normal to 8 looking into this region.
When applying (I I) to electromagnetic prob-
2x (5)

X~ ip+ —(
I) s
r)
—r
~Pe

sin OdS.
lems, wc RIc conf lontcd with gIcatcr' dlf6cultlcs
than we were when applying previous methods.
In either case, we have to make approximations
to the integrand over the surface of integration
These results agree with Sommerfcld's modi6ca- but while before we were aided in making the
tion of the Kirchhoff formula. s ncccssRI y Rppl oxllTlRtloIi 8 by physi cR1 ln tul t Ion
3, By the Induction Theorem the field E, H and experience, we have nothing to guide us, at
consisting of the reBected field E', H' inside the least at pI"cscnt, ln the case of the KirchhoR
horn and the transmitted 6eld 8",'II" outside folrnllR. . Kc couM interpret, fol cxaIHplc, V Rs R
the horn can be produced by electric and mag- cartcslRn component of thc magnetic vcctoI
netic current sheets Yivith densities given by (I). potential, but we have no knowledge of its
Khllc thc actual plane has a hole 1n It ln applying thc ' R. Darbord, L'Onde Elec. 11, 53—82 (1932).
sccoQd method wc are perInltted to plug thc hole with a H. DIamond and F. W. Dun D1orc) Nat. 8U1". Stand.
perfectly conducting disk since the 6eld due to the postu- J. Research 19, 1—19 (1937). (Paper RP 1006.) Proc.
lated current sheets is known to be identically zero on the I. R. E. 25, 1542-1560 (1937).
side of the pipe. "W. L. Barrow and F. M. Greene, Proc. I. R. E., 26,
8 A, Sornmerfeld, Gottinger Nachr. 1894, Nr. 4. 1498-1519 (1938).
 ELECTROMAGNETIC WAVES 313

behavior over the outer surface of the horn. We Equivalence Principle on one side, and from the
can expect no help from the experiment since we Kirchhoff formula on the other, becomes small;
cannot measure the magnetic vector potential this is because the "space factor" of the array of
either directly or indirectly. Likewise, we could secondary sources over the aperture becomes the
apply (11) to the electric vector potential but dominant factor and the precise nature of the
with no better success. If we simply ignore the secondary sources loses its importance. In fact,
contributions coming from the outer surface of one could make almost arbitrary assumptions
the horn, we get different radiation patterns, with regard to the secondary sources and obtain
depending upon the choice of V. If we apply (11) the radiation patterns, for large apertures, which
directly to cartesian components of E and II and would differ but little.
ignore the contributions of the outer surface of
the horn, we obtain a result which is inconsistent APPENDIx I
with Maxwell's equations. In other words, while
The essential parts of our proofs of induction
(11) is correct when applied to a closed surface, and
equivalence theorems are the appropriate
there seems to be no way to adapt it to practical
uniqueness theorems, that is, theorems asserting
needs when we are forced to make approxi-
the uniqueness of the field defined by a given set
mations. An interesting example of the difficulties
of conditions. If the given set of conditions is
which may confront us is given in Appendix III,
known to lead to a unique solution, we are
At this point a few words must be said about
assured of obtaining the same solution no matter
the comparison made by Barrow and Greene"
what method we happen to use even if we were
between experimental radiation patterns and
simply to guess the solution and then to verify it.
those calculated with the aid of the Kirchhoff
For instance, in dealing with free oscillations
formula from the magnetic vector potential. As
inside a perfectly conducting spherical sheet we
the authors point out the agreement is satis-
may obtain a certain solution subject to the
factory only in the horizontal plane (in the plane
condition that the tangential component of the
perpendicular to the electric vector). The agree-
electric intensity vanishes at the surface of the
ment would have been satisfactory in the vertical
plane (but not in the horizontal) if (11) were IOO 90 80
applied to the electric vector potential. In the
case considered by Barrow and Greene
formula (6) agrees satisfactorily with experi-
mental results in both planes. Fig. 5 represents a
comparison between an experimental vertical
radiation pattern and patterns calculated by
diA'erent methods.
180 0
I.O 0.8 0.6 04 0.2 0 0.2 0.4 0.6 0.8 I.O
RELATIVE FIELD INTENSITY
CONCLUSION
FIG. 5. (1) The experimental vertical radiation pattern
It is hardly necessary to emphasize that the obtained by Barrow and Greene for the open end of a
proposed methods of dealing with radiation and rectangular tube; this curve is taken from Fig. 108 of
reference 11; (2) the preferred theoretical curve calculated
diffraction problems are based, at least in the by Barrow and Greene by applying the Kirchhoff formula
important practical cases, on certain approxi- to a Hertzian vector; (3) the theoretical curve calculated
by Barrow and Greene by applying the Kirchhoff formula
mations. Fortunately these approximations are directly to the electric vector and disapproved by them
because the direction of the electric vector in the radiation
susceptible of direct experimental verification. It field is wrong; (4) the theoretical curve obtained from the
is hoped that this paper will stimulate such Equivalence Principle on the assumption that the ratio of
experiments.
8 to H over the aperture is the same as for plane waves
in free space; (5) the theoretical curve obtained from the
These experiments should be made preferably Equivalence Principle on the assumption that the field
over the aperture is the same as would have been over the
on fairly small apertures. As the aperture be- same surface in case the tube were continued indefinitely;
comes larger, the difference (but not the ratio) (6) the theoretical curve for the case in which a perfectly
conducting infinite flange is added to the open end of the
of the radiation intensities obtained from the tube.
314 S. A. SCHELKUNOFF
spherical sheet. The field outside the sphere may where n is the normal to S pointing away from
be taken equal to zero. Having obtained the the volume under consideration. In (15) the
solution, we find that the tangential component region v may be a multiply connected region
of the magnetic intensity is discontinuous across enclosed by surface So externally and by surfaces
the sphere. In accordance with Maxwell's equa- S&, $2, etc. internally; in this case, S consists
tions, this discontinuity implies electric current of So, Si, S2, etc.
in the sphere, the density of this current being
. In dissipative media the field at great distances
equal to the above-mentioned discontinuity. If from the sources vanishes as an exponential
we were to calculate the field of this current sheet function of the distance; hence, the last term in
by the retarded potential method, we expect to (15) vanishes in the limit as So recedes to infinity.
obtain the field from which we started. If there are no impressed currents or if in the
An exhaustive analysis of uniqueness theorems regions occupied by impressed currents, the
of electromagnetics would require considerable components of the field in the directions of these
space and could properly be regarded as a currents vanish, then the left member of (15) is
subject in itself. We shall restrict ourselves to the equal identically to zero. The first term on the
simplest case of monochromatic waves in dissi- right is real and the second is a pure imaginary;
pative media. This case is really sufficient for therefore, in absence of impressed currents these
our purposes since we can assume the conductivity terms must vanish separately. Since E E* is
of the medium to be so slight that its effect is essentially positive, the volume integral can
negligibly small and yet be sure that the needed vanish only if E vanishes everywhere; then II
theorems are applicable. also vanishes everywhere and the electromagnetic
We shall write Maxwell's equations in the field is identically zero.
following form Let us now prove with the aid of the above
result the free-space equivalence principle (4).
cui 1 E = —ice @II—3II,
(12) The assertion is that the current sheets (4) over a
curl H= (g+icue)E+ J,
surface S enclosing given sources produce the
where J
and 2II are the densities of the applied fieM E, H which is equal to the field E", H" on
electric and magnetic currents. Let us multiply the source-free side of S and to the field 0, 0 on
scalarly the first equation by the conjugate of II, the other side of S. This synthetically obtained
the conjugate of the second equation by E, and field is certainly one solution of Maxwell's
subtract; thus we obtain equations subject to the proper boundary con-
II* curl B—E H*= —3f II*—E J* ditions across S and behaving in the appropriate
curl
manner at infinity. The question is whether it is
gE E*+i(ucE—E* icvpH H~, (—
13)
the only one. In other words, shall we obtain the
where the asterisks. designate the conjugate same field E, H if we solve (12) by the retarded
complex numbers. Integrating over a volume potential method or shall we obtain a different
enclosed by a surface S and taking into account field Ej, H~& Let us suppose that the latter is the
that case. Then we have
H* curl B—Z cur/ H*=div. XXII*, curl E = —i coIJ,H —M",
curl Eg — —i~pHg —M",
= J", (16)
div. EXH*dv = (EXH*).dS, curl H (g+icue)E+
) (S) curl Hg (g+ia&e)Er+ J"—
—
.
we obtain On subtracting these equations we find that the
difference field E —E~, H —H~ satisfies the homo-
(E J*+M H*)dv=g E E"dv geneous form of (12), that is, the form in which
(e) e
(p) 3II= J=O. Therefore, this field vanishes identi-
cally throughout the entire space and E& —E,
+i(v (yH H* eE E*)dv+—
(EXII*).dS, Hl =H.
(~) (15) Similarly we can prove an equivalence
 ELECTROMAGNETIC WAVES

theorem for the case in which the horn is retained and the total transmitted voltage and current are
and the impressed current sheets (4) extend only
over the mouth of the horn. The Induction Zp Z Z Zp
I'(xp) = P, V'(xp) = U',
Theorem (1) is also proved in the same manner. Z(}+Z Z+Zp
In connection with the above theorems, the
2ZQ 2Z
reader has undoubtedly noted an interesting I"(xp) = .

IP V"(xp) = VP
paradox. The electric and magnetic current Zp+Z ZO+Z
sheets on 5 imply only discontinuities in the
magnetic and electric intensities across the These expressions are precisely the same as
surface. How is it that in accordance with the would be obtained directly from the boundary
above theorems we know the actual intensities on conditions at point x=xp.
either side of the sheet and not only their If Z is nearly equal to Zp, V' and J' are small
differences? The explanation lies in the fact that while V" and I"
are nearly equal to V' and . I
in these theorems the densities of the superficial This is the case when practically all power
current sheets are not independent, these densities carried by the wave V', I' is transmitted beyond
are obtained from functions satisfying the homo- x=xp. Similarly we expect that in the case of a
geneous Maxwell's equations and theoretically horn with a large mouth, when practically all
either the electric field alone or the magnetic power delivered to the mouth passes on into the
field alone su%ce for defining both current sheets. outer medium, E" and H" are substantially
equal to 8' and H' as we have previously assumed
APPENDIx II in our applications of the equivalence theorems.
In the case of a simple transmission line, a " On the other hand, if Z is either very small
direct verification of the Induction Theorem is compared with Zp or very large, then the
reHection is nearly complete and one of the
possible and the results are useful as an aid to the
study of more complex cases. As before we shall
quantities V", I"
nearly vanishes while the other
is nearly doubled. In this case the effect across
designate by a single prime the reHected voltage
and current and by a double prime the trans-
the impedance Z is produced largely either by the
mitted voltage arid current. The subscript ~ will series generator alone or the shunt generator
be used to designate those parts of the voltage alone. In the corresponding three-dimensional
and the current which are produced by the
case (which occurs when the frequency is near
impressed series voltage V' and the subscript 2
the cut-oR frequency) we expect, therefore, that
will refer to the voltage and current produced by
as a first approximation we can calculate the
the shunt current I'. radiation patterns by assuming only one current
We shall consider the case of a semi-infinite sheet over the mouth of the aperture. The
transmission line, with characteristic impedance corresponding radiation patterns will be sym-
metrical about the plane of the aperture of the
Zp, terminated in the impedance Z at some point
x=xp. Carrying out the calculations in accord- horn. The departure from symmetry will be
ance with the Induction Theorem, we obtain caused by the incomplete reHection.

Ir'(xp) = UP/(Zp+Z), APPENDIx III

Ip'(xp) = ZP/(Zp+Z), —
Ir" (xp) = V'/(Zp+Z), Consider a perfectly conducting semi-infinite
Ip" (xp) =ZpP/(Zp+Z), coaxial pair with an open end. Let the radii a and
r'(xp) = —Zp V'/(Zp+Z), (17) b(b)a) be very small compared with the wave-
Up'(xp) =ZpZP/(Zp+Z),
length. The field inside the coaxial pair is
Vg" (xp) =ZV'/(Zp+Z), approximately
Vp" (xp) =ZpZP/(Zp+Z). 60I 27r

Thus, the total reHected voltage and current

cos Ps, Hp= sin Ps, P= —,(1)
p 2' p
"A transmission line admitting of only one transmission
111ode. where: I is the maximum amplitude of the
316 S. A. SCH ELKUNOF F

iI
2, = ——(b'log ——a' log —
current, X the wave-length, p the distance from b aq
the axis and 2' the distance along the axis. The ~ ~

4)
open end is in the plane s = 0.
E. pp po&
- —iver
The field can be obtained from a vector i (b2
~
n2)
potential whose components are
iI
—log —
p
A, =A„=O, A, = sin Ps. (2) From this we calculate the field and then the
2' pp radiated power; thus we obtain
This can be verified by substitution in 40~'
W=
p
b' log -—
—
b
a' log —
II„= —,
gA,
Ep ——
zcoE'
1 c}'A,
8 pBz
i
Po po&
i

—k(b' —a') I'. (&)

The values of the above wave function and its
normal derivative over the aperture are
By choosing po properly we can make the radi-
ated power W equal to anything from zero to
=—
BA, ip
A, =O, I log —. p
infinity. The approximate value for W found
&0 2T pp
with the aid of the Equivalence Principle corre-
Using the Kirchhoff formula in the customary sponds to pp satisfying
manner (i.e. , by applying it to the aperture) we 6
find that at great distances from the aperture the 5' log —
b
—a' log — = 0.
approximate value of A.- is Po po

Note: On Diffraction and Radiation of Electromagnetic Waves

J. A. STRATTQN AND L. J. CHU
Massachusetts Institute of Technology, Cambridge, Massachusetts

Dr. Schelkunoff has kindly shown us the manuscript of the preceding paper
which is closely allied to ours. "
Since our formulation of the problem differs
somewhat from that of Dr. Schelkunoff in "Some Equivalence Theorems of
Electromagnetics and Their Application to Radiation Problems, and in ""
the preceding paper, it appears worth while to point out that the results ob-
tained by the Equivalence Principle are identical with ours. If the field of the
equivalent surface currents is calculated from the vector potential or the
Hertz vector, the contour charges need not be introduced explicitly. The
contour charges simply ensure the self-consistency of the assumed field on
the surface.

"J. A.99Stratton and L. J. Chu, "Diffraction Theory of Electromagnetic %aves, " Phys,
Rev. —107 (1939).
"S. A. Schelkunoff, Bell Sys. Tech. J. 15, 92—112 (1936).
SO,
",