a

r
X
i
v
:
p
h
y
s
i
c
s
/
0
4
0
1
0
6
7

v
1



1
3

J
a
n

2
0
0
4
Expressing Maxwells equations independently of
the unit systems
C. Vrejoiu
Facultatea de Fizica, Universitatea din Bucuresti, 76900, Bucuresti, Rom ania
E-mail : [email protected]
Abstract. A procedure to teach Electrodynamics independently of unit systems is
presented and compared with some of those given in physics literature.
1. Introduction
As it is remarked in [1], it is a well-known fact that one major hurdle for students in a
class of electromagnetism is to get familiar with the adopted unit system, and to move
from one unit system to another (e.g. SI to Gaussian). As a student and, later, as
an Electrodynamics professor, I have felt myself this hurdle. However, inspired by the
Jacksons book [2], I have adopted a procedure to teach electrodynamics independently
of unit systems since 1985 and used it consistently in my lectures [3]. Here is a summary,
including some comments regarding results from [1, 2] and some inconsistencies from
[3].
2. Writing the Maxwells equations
Before writing the Maxwells equations we have to dene the physical system itself,
namely the electromagnetic eld (EMF), by the interactions with other known systems.
The Lorentz force may be experienced for introducing the electric charge q and the
fundamental variables E and B of the EMF. The electric eld E is dened by the force
acting on a charge at rest in the inertial system of the laboratory. The relation between
force, charge and electric eld, as is considered in all known unit systems [2] is
F = qE,
such that the introduction of an arbitrary proportional constant in this relation is not
of practical interest. Dierent unit systems are introduced when expressing the Lorentz
force acting on a moving charge q
F = qE +
q

0
v B, (1)
In [3] the magnetic eld is dened within the old formalism of the magnetic shells or sheets and, the
constant
0
is introduced by the equation of the torque on an elementary shell C = (1/
0
)IS B =
mB where the magnetic moment is m= (1/
0
)IS. From this torque one may deduce the Laplaces
Expressing Maxwells equations independently of the unit systems 2
Here, the constant
0
depends on the unit system used.
The law of electric ux introduces a second constant k
e
:
_
D
E dS = k
e
Q(D)
or in the local form
E = k
e
.
Amp`eres law introduces a third constant k
m
_

B dl = k
m
I(

) or B = k
m
j.
Maxwells equation that generalizes the Amp`eres law is
B = k
m
_
j +
1
k
e
E
t
_
.
Lets adopt, for practical reasons, the following new notations:
k
e
=
1

0
, k
m
=

0

0
.
Here, the constants
0
and
0
are proportional constants, their values depending on the
units used. The purpose of such denitions is to write the laws of EM with notations
specic to the international unit system (SI).
The law of electromagnetic induction is written as
_

E dl =
1

i
_

B
t
dS or E =
1

i
B
t
,
i
> 0
where
i
is the last constant introduced. Hence, the Maxwells equations are
1

0
B =
1

0
j +

0

0
E
t
, (2)
E =
1

i
B
t
, (3)
B = 0, (4)
E =
1

0
. (5)
force (up to a gradient) and from this force, the presence of the factor 1/
0
in the Lorentz force
expression.
In [3] the constant
0
is introduced dening the magnetic scalar potential of an elementary magnetic
shell
(r) =

0
4
m r
r
3
.
Consequently, the factor
0
occurs in the Amperes law.
Expressing Maxwells equations independently of the unit systems 3
Several arguments justify the equality between
0
and
i
. One may argue that

i
=
0
as in Jacksons book [2], but we may use directly the equation (1) considering a closed
conductor moving with a constant velocity u in an external nonuniform magnetic eld
B(r) in the laboratory system L. The electromotive force corresponding to the electric
current due to the Lorentz force is [4]
E =
1

0
_

(u B) dl =
1

0
1
t
_

(dl a) B =
1

0
N
m
t
where a is the displacement vector of the element dl of the current contour during the
time t, and N
m
is the variation of the magnetic ux N
m
through a surface attached
to this contour due to the displacement a. After comparing the last equation to the
equation (3), and after considering that only the magnetic ux variation is the dening
element of the electromagnetic induction, we have to admit the equality between the
two constants
i
and
0
.
In [2] this equality is argued theoretically using the Galilei invariance of Maxwells
equations for u << c: for the observer in the system L

of the conductor the eect is

associated with an induced electric eld
E

=
1

0
u B, (E = 0).
This equation, together with the transformation law B

= B, is, indeed, the rst

approximation of the relativistic transformation law.
Another argument for considering the two constants equal is given by the physical
requirements of the EM theory. The denitions of charge, energy etc must complete
the Maxwells equations together with the corresponding theorems resulting from these
equations. Combining the equations (2) and (3) one obtains
1

0
B (B) +
0
E (E) =
1

0
j B

0

0
E
t
B

0

i
E
B
t
and, nally,

0
_
1

0
E
t
B +
1

i
E
B
t
_
= T f (6)
where f = E + (1/
0
)j B is the Lorentz force density, and
T
ik
=
1
2
_

0
E
2
+
1

0
B
2
_

ik

_

0
E
i
E
k
+
1

0
B
i
B
k
_
.
The equation (6) represents the relation between the electromagnetic forces and the
Maxwells stress tensor T
ik
= T
ik
within the static case. It can be considered as the
Expressing Maxwells equations independently of the unit systems 4
EM momentum theorem if the left hand term is a time derivative. Consequently, this
requires
i
=
0
and denes the electromagnetic momentum density by
g
em
=

0

0
(E B) .
Lets consider
i
=
0
from now on. We point out that the equations are written
using SI units by substituting
0
= 1 ,
0
=
0
,
0
=
0
.
In the case of the free EMF ( = 0, j = 0) one obtains the propagation equations
for E and B,
E

0

0

2
0

2
E
t
2
= 0, B

0

0

2
0

2
B
t
2
= 0
From the last equations one obtains the fundamental relation

2
0

0

0
= c
2
, (7)
between the three constants
0
,
0
,
0
introduced in the Maxwells equations and an
experimental constant, the light speed c. The two remaining arbitrary constants dene
dierent unit systems.
The electromagnetic potentials are introduced by the equations
B = A, E =
1

0
A
t
and verify
A

0

0

2
0

2
A
t
2
=

0

0
j +.
_
A+

0

0

t
_
,
=
1

0

1

t
A.
The gauge transformations are
A A+,
1

t
,
and the Lorenz constraint is
A+

0

0

t
= 0.
Correspondingly, the equations of the potentials in this gauge are
A
1
c
2

2
A
t
2
=

0

0
j,
1
c
2

t
2
=
1

0

with the retarded solutions

A(r, t) =

0
4
0
_
j(r

, t R/c)
R
d
3
x

, (r, t) =
1
4
0
_
(r

, t R/c)
R
d
3
x

. (8)
Those of my students who do not agree with my general notation are free to use this choice and, so,
to work in SI units. They are notied on this freedom from the rst class.
Expressing Maxwells equations independently of the unit systems 5
The above notations are specic for rationalized unit systems. To nd out the
changes required to rewrite the Maxwells equations using non rationalized unit systems,
lets consider the solutions (8) for the scalar and vector potentials. We convert the
Maxwells equations (2)-(5) to non rationalized notations by eliminating the factor 1/4
from (8)

0


0
4
,
0
4
0
.
The equation (7) is invariant to these transformations. The Maxwells equations with
the notations of a non rationalized unit system are
1

0
B =
4

0
j +

0

0
E
t
,
E =
1

0
B
t
,
B = 0,
E =
4

0
.
The Maxwells equations (2)-(5) with
i
=
0
and all their consequences are
rewritten in all usual unit systems by substituting the following values for the three
constants:
SI:
0
= 1,
0
=
0
,
0
=
0
,
Heaviside:
0
= c,
0
=
0
= 1,
Gauss:
0
= c,
0
=
1
4
,
0
= 4,
esu:
0
= 1,
0
=
1
4
,
0
=
4
c
2
,
emu:
0
= 1,
0
=
1
4c
2
,
0
= 1.
These notations work also within the relativistic electrodynamics (in vacuum). The
relativistic equations of motion of a charged particle are obtained from the Lagrange
function
L(t) = m
0
c
2

1
v
2
c
2
q +
1

0
qv A
which may be written with an invariant parameterization as
L() = m
0
c
_
x

() x

()
1

0
A

(x) x

where
(A

) =
_

0
c
, A
_
.
is the 4-potential. The relativistic invariance of the motion equations and of the
Maxwells equations is realized by dening T

as components of a
tensor. In particular, A

may be considered as the components of a 4-vector although

Expressing Maxwells equations independently of the unit systems 6
this is not a necessary condition of the invariance of the theory, except the request to
have covariant equations for the 4-potential.
As it is pointed out in [2, 1], many diculties are encountered trying to generalize
this procedure to the macroscopic electromagnetic eld in the presence of a medium.
The complications arise due to some inconsistencies in the denitions adopted in various
unit systems as it is pointed out in [1].
In [3] one denes the vectors D and H by the equations
D =
0
E +P, H =
1

0
B M
such that one may obtain the macroscopic Maxwells equations only in SI and Heaviside
unit systems. In [1], the denitions of D and H are given by introducing two new
constants
d
and
h
(in [1] labeled k
D
and k
H
). With our notations, they are
D =
d
_
E +
1

0
P
_
,
H =
h
(B
0
M) .
However, in [1] a new constant
m
(in [1] labeled k
M
) is introduced by the relation
between the magnetization current j
m
and the magnetization vector M
j
m
=
m
M.
This new constant is not necessary. Actually, we can reduce the number of
supplementary constants to two, as in [2].
The constant
m
is necessary in the denition of the magnetic dipolar moment
(and in all multipolar orders). If we consider the Laplaces force, the corresponding
expression is well dened by the Lorentz force. Furthemore, for a steady current loop
we have
F(I, ) =
I

0
_

dl B
We may demonstrate the relation
_

dl B =
_
()
(n )BdS.
So, the Laplaces force is equivalent, at least regarding the resulting force, to a ctitious
force acting on the shell ()
dF

=
I

0
dS.
The simplest and natural denition of the magnetic moment dm corresponding to an
elementary shell is
dm=
I

0
dS.
Curiously, in a considerable part of the physics literature, the vectorial character of the 4-potential
is presented as a necessary condition for the relativistic invariance of the theory.
Expressing Maxwells equations independently of the unit systems 7
Also, it is possible to relate the torque dmB to the Laplaces force.[3]. Therefore, in
the case of a current distribution j in D the magnetic dipolar moment is dened by
m =
1
2
0
_
D
r jd
3
x.
Generally, one may dene the nth order magnetic moment by the tensor [5]
M
(n)
=
n
(n + 1)
0
_
D
r
n
j d
3
x. (9)
The magnetization current j
m
is given by the relation
j
m
=
0
M
where M includes the contributions of all magnetic multipoles. This is a result of the
denition (9) and of the average of microscopic equations of EMF. In conclusion, the
equality

m
=
0
is justied while a third supplementary constant is not necessary in the case of the
macroscopic eld.
3. Conclusion
By writing the macroscopic Maxwells equations as in [3],
H =
1

0
j +
1

0
D
t
,
E =
1

0
B
t
,
B = 0,
D =
only the equations in SI and Heavisides systems are obtained. To change the unit
system from Heavisides to the Gaussian one, we have to memorize some factors 4.
With notations from the present paper we have
H =

0

0
j +

0

d
1
c
2
D
t
,
E =
1

0
B
t
,
B = 0,
D =

d

0
, (10)
and
D =
d
_
E +
1

0
P
_
, H =
h
(B
0
M) . (11)
Expressing Maxwells equations independently of the unit systems 8
The equations (10) result from the microscopic equations using the relations
<
micro
>= P, < j
micro
>= j +
0
M +
P
t
,
and the denitions (11). The following values for the two supplementary constants,
named in [1] conventional constants, should be substituted to obtain the equations
within dierent unit systems. For the two supplementary constants in the usual unit
we have the following expressions:
SI:
d
=
0
,
h
=
1

0
,
Heaviside :
d
= 1,
h
= 1,
Gaussian :
d
= 1,
h
= 1,
esu :
d
= 1,
h
= c
2
,
emu :
d
=
1
c
2
,
h
= 1
Although the number of the conventional constants is reduced to two, the conclusion
from [1] remains valid: the complications due to these conventional constants make the
result not as appealing as that obtained in the vacuum case....
Acknowledgments
I thank Dr Sorina Zota for reading the manuscript and for many helpful comments.
References
[1] Leung P T 2004 A note on the system-free expressions of Maxwells equations, Eur. J. Phys. 25
(online at stacks.iop.org/EJP/25/N1)
[2] Jackson J D 1975 Classical Electrodynamics (Wiley New York) p.755-8
[3] Vrejoiu C 1993 Electrodynamics and Relativity Theory (in romanian) EDP
[4] Tamm I E 1979 Fundamentals of the Theory of Electricity (Mir Publishers Moscow)
[5] Castellanos A, Panizo M, Rivas J 1978 Am.J.Phys., 46 1116-17