A New Kind of Parity and Loschmidt’s Paradox

Fátima Masot-Conde
fatima@us.es

ABSTRACT
The conventional parity operation consists of the exchange of ―right‖ and ―left‖, hence
its alternative name, ―mirror inversion‖, since it is the one performed by common plane
mirrors. Such specificity in the inversion surface, though, is pointed out in this paper as
being responsible for its less fundamental character. Here, a new kind of parity is
introduced, whose inversion surface is a general, closed one, and consists of the
exchange of the inside and outside subspaces that the surface defines. It has been named
―inside-out‖ inversion, or simply, space inversion, and encloses the conventional right-
left parity as the particular case when the inside and outside are identical. As an
example, Maxwell‘s equations in free space will be considered. This case is important
since the time-reversible character of classical electromagnetism is under question due
to the irreversible physical behaviour of radiation. For this case, both space and time
inversions will be carried out and compared, and it will be shown that the
electromagnetic field in free space is space-invariant, while it is not time-invariant, as
commonly assumed.

Keywords: Radiation arrow, Parity, Time reversal, Maxwell‘s equations, Magnetic

Monopole

PACs: 03.50.De, 03.50.-z

1 Introduction
1.1 Loschmidt’s paradox
2 Space symmetry in Maxwell’s equations
3 Time symmetry in Maxwell’s equations
3.1 Comparison between t-backwards and t-forwards: the double symmetry
3.2 Electromagnetic plane waves at a temporal boundary
3.2.1 Exchange in handedness
3.2.2 Exchange in fields
3.3 Comparison of space and time inversions
4 Four dimensional formulation
5 Discussion of the solution
6 But what is a time border?
6.1 The absolute zero-point temperature
6.2 The Big Bang
6.3 Remarks on Energy

1
7 Conclusions
Appendix I: Space reversibility of Ampère-Maxwell’s equation and the continuity
equation
Appendix II: Handedness in the vector product
Appendix III: Notation for the four dimensional formulation
Appendix IV: Further comments on Energy

1. Introduction
The conventional parity conservation law, or mirror symmetry, used to establish —
wrongly— that Nature is not able to distinguish the real world from its image in a
mirror [1]. With hindsight, we realize the fragility of that principle. If we asked what
kind of mirror would be able to perform that fundamental exchange, the answer would
be ―the plane ones‖. Such specificity is the hint that the mirror inversion was not so
fundamental, after all. Why precisely plane mirrors?

Indeed, the inversion surface for the right-left parity is a plane. But perfect plane
surfaces do not exist in Nature. The kind of limitation in having a perfect plane is not
only circumstantial, because of the technological impossibility of making perfect
planes, without imperfections. Beyond that, the impossibility comes from a natural limit
imposed by the curvature of space-time itself: It is well known that, despite the fact that
space-time is almost euclidean (plane) everywhere, it suffers a bend in the vicinity of a
black hole and/or a singularity (in general, in the vicinity of mass). The space-time
bending is a fundamental limit in the possibilities of making, —and defining—, a plane
surface at determined quantum and cosmic scales. The non-fundamental nature of the
conventional parity becomes obvious in those regions where perfect zero curvature
surfaces are not (and cannot) be defined.

Obviously, it can be replied that the ―plane‖ for the mirror inversion is not a physical,
but a mathematical object. Taken as such, a plane is geometrically a spherical surface of
infinite radius, and hence, it is a very particular surface within a kind (of closed
surfaces, in general), and the inversion with regard to it, a very particular inversion
within a kind. Such a degree of specificity contrasts with the universality of natural
laws. Why is so much specificity demanded for laws which, on the other hand, are
supposed to be universal?

This paper introduces an extended concept of parity, which precisely relaxes the
specificity of the conventional parity transformation. The new parity consists of the
generalization of the inversion surface, now not a plane, but a general, closed one. The
new parity establishes the symmetry of natural laws with regard to an exchange of the
inside and outside subspaces that the surface defines. This generalization is based on the
idea that a natural law should not be spatially bound: 1) its range of validity should be
universal, not dying or ending at the boundary of an arbitrary region, and 2) the
arbitrary surface should not modify the law, i.e., the law should not distinguish between

2
regions (otherwise, its validity would be spatially bound, against the premise). These
conditions force the symmetry of the law with regard to any possible point of view of
the observer: for any closed surface, natural laws should be symmetric (invariant) under
the exchange of inside and outside. A good name for this symmetry could be ―inside-
out‖ parity, or simply, space parity.

The generalization of the inversion surface is not a mere generalization for no apparent
reason. It is precisely motivated by the case of classical electromagnetism, since a
closed surface is a geometrical boundary, able to enclose emitters (or receivers) of
electromagnetic waves. Once emitted, the field crosses the surface in one determined
direction: For the emitter inside (the receiver outside), the field exits, and reciprocally,
for the receiver inside (the emitter outside), the field enters. Whichever is inside or
outside, the reciprocity between emitter-receiver is the same as the one between inside-
outside, disregarding their specific locations. The exchangeability of spaces is the key to
the spatial reversibility of the electromagnetic flow. However, the field is not able to
spontaneously change its flow-sense, as would be required for the field to be time-
reversible1. The aim of this paper is to extract that conclusion from Maxwell‘s
equations, against the orthodox view which defends the theoretical time reversibility of
the electromagnetic field. But first, let us put in context that problem of classical
electromagnetism with time-inversion.

1.1 Loschmidt’s paradox

It is well known that electromagnetic fields diverge from accelerated charges, while the
opposite phenomenon of converging fields into accelerated charges has never been
observed. This arrow of radiation, as it is called, seems to collapse with the apparent
time-symmetry of Maxwell‘s equations, commonly taken as fully time-reversible, since
they do not seem to differentiate between retarded and advanced solutions. Nature,
indeed, prefers the retarded solution over the advanced one, an asymmetry that does not
fit into the proclaimed time-symmetry of Maxwell‘s equations.

This disparity between theory and reality is called ―Loschmidt‘s paradox‖ [2]. Many
authors have studied the question of where the time arrow is filtered into our
electrodynamics equations. One of the most popular and well-known answers is the
Wheeler-Feynman theory of infinite absorbers [3]. This theory postulates the need for
―absorbers‖, on one hand, (in order to cancel the advanced solution), which, on the
other hand, need to be ―infinite‖, (in order to avoid the infinite contribution of the self-
field of the accelerated charge to its own radiation, saving the renormalization problems
associated with it). The problems with this theory are diverse: one is whether absorbers
are really temporal inverters of emissions, a hypothesis which is needed so that the
advanced solution can be perfectly cancelled. Also, there is the question of what
happens in absence of absorbers (and sources), for instance, when the electromagnetic
field propagates in free space, and why it behaves there as fully retarded. Another
version of this theory was given by H. Price. In his book Time‘s arrow and Archimedes‘
Point and previous articles [4], he is convinced that it is just a matter of scale: the
incoherence of the infinite absorbers at macroscopic level is just what makes the perfect

1
That would require not a simple relocation emitter-receiver, but a transmutation of an emitter into a
receiver and vice versa, with their locations unchanged.

3
cancellation of the advanced solution, —which for him actually exists—, impossible.
Thus, the perfect symmetry on a microscopic scale, in his opinion, holds true, despite
the macroscopic radiation arrow on a large scale. Another author, D. Zeh, insists on the
absorbers idea, despite linking the definition of the ideal absorber to past macroscopic
boundary conditions of space-time [5].Following this line of argument is the article by
J. North, who links the arrow of radiation to the macroscopic thermodynamic arrow of
time [6]. With all due respect, this idea seems to me a bit dodgy: It deliberately places
the origin of the macroscopic arrow far enough, so as not to have to worry about it: the
unreachable (and untestable) initial conditions of the Universe —the so-called ―Past
Hypothesis‖—. Approaches like these leave us with a strange feeling of dissatisfaction:
their vagueness seems to be an emergency solution, rather than a real explanation in
search of a genuine understanding of the matter. But even accepting this weak
explanation as the origin of the asymmetry of radiation, a question is still left
unanswered: if the behaviour of radiation is asymmetric, why then is classical
electromagnetism time-symmetric?

It is remarkable that all the above-mentioned authors, in agreement with classic

interpretations [7-8] have accepted beyond any doubt that classical electromagnetism is
time-reversible, because of the fact that Maxwell‘s equations ―do not differentiate‖
between retarded and advanced solutions. With the exception of D. Albert [9], the
possibility that Maxwell‘s equations are actually not reversible in time has never been
suggested. Even a different view, put forward by M. Frisch, maintains, like all the
previous authors, that Maxwell‘s equations are time-symmetric, and in order to match
the observed asymmetry of radiation with the apparent symmetry of equations, he needs
to ―delimit what is physically possible‖, by means of adding an additional condition, the
retardation condition, on a basis of causality and real physical phenomenology [10].

In this paper, I question the main premise that classical electromagnetism is time-
invariant, as commonly accepted. More specifically, the aim of this paper is to show
that Maxwell‘s equations in free space are not. But first, we need to understand what
kind of reversibility they do fulfil. Starting with Maxwell‘s static equations, this is dealt
with in the next section.

2 Space symmetry in Maxwell’s equations

The conventional parity is the type of inversion performed by a plane mirror. Here we
are interested in relaxing the inversion surface, to a general closed one, in order to
define a more general parity concept. The idea is simply like turning a sock inside out:
for a given surface, the space inversion consists of the exchange of one subspace for the
other, the inside is treated as it was previously the outside, and vice versa. A function
would be space-invariant if it stood unaltered after the space exchange. As the plane is a
particular case of a closed surface, —which closes at the infinite—, it is obvious that the
right-left parity is a particular case of space parity, where both subspaces are identical.
For that reason, if a law is space-invariant, it must also be right-left invariant, but the
opposite, in general, is not true.

Due to its generality, the application of this kind of inversion may not be so
straightforward as the conventional one, unless the surface is explicitly involved in the

4
laws. That is the case of classical electromagnetism, at least with Gauss‘s equations,
which are written in integral form in terms of an arbitrary closed surface S:

q
 S
E  dS =
0
(1.a)

 S
B  dS = 0 (1.b)

where q is the charge enclosed within, and dS is the surface vector pointing
outwards, (see figure 1).

Figure 1: Conventional definition of the surface and circulation vectors at each point of a
closed surface. The circulation vector d is contouring C in the indicated direction,
according to the right-hand rule.

Let us distinguish with subscripts the two subspaces defined by the surface: subspace I
is the inside, (shadowed) and subspace O is the outside, (white). Let us call the charge
within subspace I qI , and correspondingly, the charge which belongs to subspace O,
qO . These are not independent: in this paper, both the charge conservation principle
(i.e., the charge is neither created nor destroyed at any point of the Universe) and the
charge neutrality of the Universe2 are assumed. This means that, for any charge isolated
within a surface, there must be the same amount of opposite charge somewhere outside
the surface. Mathematically:

qI   qO , (2)

As we are interested in checking the reversibility of equations (1) with regard to the
exchange of both subspaces, let us place two observers, one in the subspace I (observer
I) and the other in the subspace O (observer O). Both are willing to formulate
Maxwell‘s equations, for which they should adopt some kind of convention for defining
the surface and circulation vectors. If they, for example, agreed in taking the surface

2
The charge conservation principle establishes the constancy of the total net charge of the Universe,
although its value is not specified. In this paper I follow the authors in [11-12] in taking it as zero,
meaning the balance of positive and negative charge.

5
vector dS as pointing ―outwards‖, and defining the corresponding circulation vector at
each point d as given by the right-hand rule, observer-I, from his point of view,
would take this couple of vectors as being dS I and d I (figure 2.a), while observer-O,
from his, would say that they are dSO and d O (figure 2.b).

As the outside of one is the inside of the other, note that

dSI   dSO (3)

And the circulation vectors also verify:

d I  d O (4)

With this notation, Gauss‘s electric equation (1.a) takes the form:

qI
 S
E  dS I =
0
(5)

By using equations (2) and (3), in (5):

qI qO
 S
E  dSI =  S
E  (  dSO ) 
0

0
(6)

This leads to Gauss‘s complementary law expression in terms of the O subspace:

qO
 S
E  dSO =
0
(7)

Note that it is formally identical to (5), meaning the symmetry of the law with regard to
the inside-out exchange3.

For obvious reasons, Gauss‘s magnetic law (1.b) is even more clearly symmetrical with
regard to the space exchange. Let us now check the space inversion for Ampère and
Faraday‘s laws.

For the case of Maxwell‘s time-dependent equations, only propagation in free space will
be taken here into consideration, since radiation —whose time-behaviour is our goal—,
is normally assumed as the specific case of the unguided transport of waves and energy
through empty space. For reasons that will be apparent in the next section, the free-
source terms have been here omitted on purpose, for the sake of simplicity, and also to

3
It is remarkable that the hypothesis of space symmetry of Maxwell‘s equations has to assume the
hypothesis of charge neutrality of the Universe. They need each other: reciprocally, the space symmetry
of Maxwell‘s equations could have been assumed beforehand, and taken to support the hypothesis of the
charge neutrality of the Universe.

6
avoid contamination or mistakes coming from the conventional approach, (for the
interested reader, the space reversibility of Ampère-Maxwell‘s, and the continuity
equation, can be found in Appendix I). In free space, this is the way in which Ampère
and Faraday‘s laws are written in integral form (natural units, the light velocity c  1):

  E  dS
 C
Bd = S
t
(8.a)

  B  dS
 C
E d =  S
t
(8.b)

The surface S in these equations is now open, (limited by the contour C, whose
direction is linked to the surface vector according to the right-hand rule). Still, we can
always take it as a part of a bigger, closed surface (a window). With this borrowed
interpretation from the previous case, it makes sense to keep calling the orientation of
dS as either inwards dSO , or outwards, dS I , the ―cavity‖, (despite S, the window
surface, being open), each one linked to its own circulating vector, d I and d O. Like
the previous ones, they also verify equations (3) and (4).

Figure 2: Definition of the surface and circulation vectors for (a) the inside (I-) and (b)
outside (O-) subspaces. The circulation vectors d I and d O are respectively
contouring CI and CO. Note that the I-vectors coincide with the conventional definition
in figure 1.

With this notation, Ampère‘s law (8.a) can be rewritten in terms of the I-subspace as
(natural units):
  E  dS I
 C
Bd I = S
t
(9)

7
The question is whether or not we can write a similar equation in terms of the O-
subspace. The answer is yes. By using equations (3) and (4), it is obvious that:

  E  dSI   E  (  dSO )
S
 S
(10)
t t
And

 C
B d I =  C
B  ( d O ) (11)

Therefore, an identical expression to (9) can be obtained for Ampère‘s law in terms of
the O-subspace:

  E  dSO
 C
Bd O = S
t
(12)

The same derivation can be made for Faraday‘s law (8.b). Maxwell‘s free-space
equations, to sum up, stand on a spatial boundary S:

Spatial Boundary (I  O)

Subspace I Subspace O
  E  dS   E  dS
 
I O
Bd = S
Bd = S
C I
t C O
t

  B  dS   B  dS
 
I O
E d =  S
Ed =  S
C I
t C O
t

Note that this kind of inversion is spatial (time has been untouched in it). The symmetry
of the laws relative to the space-exchange is obvious: The electromagnetic field, such as
we know it, can either enter or exit a closed surface, since the equations that govern it do
not distinguish between inside and outside. This, which may seem a triviality, will not
seem so when compared with the different behaviour of the electromagnetic field at a
temporal boundary. This will be dealt with in the next section.

8
3. Time symmetry in Maxwell’s equations

Let me start this section with an inspiring paragraph by Sakurai [13]:

In this section, we study another discrete symmetry operator, called time

reversal. This is a difficult topic for the novice, partly because the term
time reversal is a misnomer (…). Actually, what we do in this section can
be more appropriately characterized by the term reversal of motion.
Indeed, that is the terminology used by E. Wigner, who formulated time
reversal in a very fundamental paper written in 1932.

The paper Sakurai is talking about is [14].This piece suggests that the electromagnetic
field has been commonly viewed in purely mechanicistic terms: its behaviour in time is
subordinated to the behaviour of the charge in motion [15].

This subordination implies a particular way to solve Maxwell‘s equations with the time
backwards. That way consists in reassigning the minus coming from the time backwards
to one of the fields, —which usually is the magnetic field [16]. So the approach in the
textbooks for the time inversion lies in the substitution of B by –B in Maxwell‘s
equations, considering the effect that a charge moving backwards would produce on the
magnetic field. Still, there is another approach by Feynman, which, instead, reassigns
the minus coming from the time backwards to the electric field, (i.e., E is replaced by –
E in Maxwell‘s equations), now considering the cause, i.e., the fact that, for a charge to
reverse its motion, it would require a reversed electric field. Both approaches are
complementary and mutually-excluding. This exclusion is the seed of their
incorrectness: both are possibly right, so none of them can be, since the one‘s success
implies the failure of the other. Correctness is compromised in the dilemma of choosing
between cause or effect, indissolubly connected by the time-link we are looking for.

Leaving interpretations aside, if for the time reversal we need to take on board B by –B (or
E by –E) in Maxwell‗s equations, this means that, strictly speaking, we are not solving
Maxwell‗s equations: we are solving Maxwell‗s equations plus a condition in the field. We
are therefore driving the solution, on the basis of what we think the time reversal might be,
—a perception that may be wrong if the time reversal is more extraordinary than
expected.—.

In either case, Maxwell‘s equations with the time backwards become identical to the
time forwards set, (and their solutions, obviously, too), so the electromagnetic field with
the time backwards turns out to be the same as the one with the time forwards, (except
for a phase shift), which makes inexplicable why then it is not observed, if it is so real.

These conventional approaches are based on that single principle, —I would rather say a
prejudice—, which considers matter the prime reality, and the field, secondary (in
Wigner‘s own words [1]). At least we have to concede that there exists another way to
focus the problem: to invert the importance of matter and field, and consider matter as
the subordinated reality, and the field, the prime one. That is the new approach explored
in this paper.

9
3.1 Comparison between time-backwards and forwards: The double symmetry

If matter is not relevant for learning the time behaviour of radiation, we can take it away
from the equations, and restrict our analysis to Maxwell‘s equations in empty space.
Time running forwards makes the passing of time t positive, and running backwards,
negative. This has been taken into account with the following notation:

t t  0
t   (13)
  t t  0

For a better comparison, let us write both Maxwell‘s sets, forwards and backwards,
together. In terms of the I-subspace4, since it is identical to the convention, and in
natural units, (the light velocity c=1):

Temporal Boundary (t  t )
t (forwards)   t (backwards)
  E  dS   E  dS
 
I I
Bd = S
(14.a) Bd =  S
(15.a)
C I
 t C I
 t

  B  dS   B  dS
 
I I
E d =  S
(14.b) Ed = S
(15.b)
C I
 t C I
 t

Set (14) in the left is the usual t -forward, while set (15) in the right is the time-inverted
set. We can get rid of the explicit minus in equations (14.b) and (15.a) by means of
equation (3):

t (forwards)   t (backwards)
  E  dS   E  dS
 
I O
Bd = S
(16.a) Bd = S
(17.a)
C I
 t C I
 t

  B  dS   B  dS
 
O I
E d = S
(16.b) Ed = S
(17.b)
C I
 t C I
 t

4
It is not necessary the inside-out notation in this case, (see for instance [15], in terms of right and left). I
have used it here for keeping the parallelism with the space inversion, and highlighting the difference
between both cases.

10
Clearly, equations sets (16) and (17) are of opposite handedness: In (16.a), d I is
linked to dS I , so it is right-handed (surface and contour are linked by the right-hand
rule). In (16.b), d I is linked to dSO , thus it is left-handed (surface and contour are
linked by the left-hand rule). Set (17) is the mirrored set (16), if we just allow a change
of handedness in the contour-surface link: equations (16.a) and (17.a) are respectively
the right and left versions of the same equation, and so are equations (16.b) and (17.b),
the left and right versions of the same relation between E and B. Actually the four are
all the very same equation, if we just allow the exchange of handedness
Right  Left and fields E  B .

t (forwards)   t (backwards)
Handedness Symmetry R  L

Right-handed Left-handed

Left-handed Right-handed
Field Symmetry E  B

This way of writing both sets reveals the hidden symmetry between them. Neither of
them seems more special than the other. If we accept no conditions in the fields for set
(16), we have no right to deny the same status (condition-free) to set (17), or otherwise,
we would be deliberately breaking the symmetry, by favouring one set over the other.

Now both sets are not reversible, against the general opinion, but the symmetry between
them is double: one Maxwell‘s set becomes the other either if we perform an exchange
of handedness, (here called ―H-symmetry‖), or equivalently, an exchange of fields (―F-
symmetry‖). By means of simple algebraic relationships, both transformations can be
checked to be equivalent. By direct comparison with the time-forwards wave, the time-
backwards wave can be easily obtained. Obviously, the two waves must reproduce the
same double symmetry of the equations they come from. For the simplest case of a
plane wave, this is illustrated in the next section.

3.2 Electromagnetic Plane Waves at a Temporal Boundary

It has been shown the double time-symmetry of Maxwell‘s equations. Let us explore
now the solutions, considering the simplest far-field case in lossless media, where the
field becomes a plane wave and Maxwell‘s equations turn into a simple set of algebraic
equations. As a reference, let us recall the usual time-forward set, which is written in
this case as follows (natural units):

11
 t (forwards) 
kˆ  E=0 kˆ  E=B
(18)
kˆ  B=0 kˆ  B=  E
E
where  1, and the cross product is given, as usual, by the right hand rule.
B

3.2.1 Exchange of handedness

If the exchange of handedness is performed in equations (18), the t-backward set is

obtained. For a better comparison, let us write them together:

Temporal Boundary
t (forwards)   t (backwards)
kˆ +  E + = 0 kˆ   E  = 0
k̂ +  B + = 0 kˆ  B = 0
 

kˆ  E = B
+ R +  (19) kˆ  L E  = B  (20)
k̂ + R B + =  E kˆ  B =  E
 L  

The subscripts + and - underline the domains, t-forwards or backwards, for which they
are defined Here, again, E  1, for both (19) and (20), and k̂ + and k̂  are the direction
B
vectors of the t  and t  waves, with

kˆ +  kˆ  , (21)

so that the t  and t  waves have opposite directions.

Note that the handedness is only involved in the cross products, so they now need to be
distinguished,  R right- and L left-handed. For set (19), the right-hand is usually
adopted, and so it has been kept here, —the cross product R in (19) is identical to the
usual cross product  in (18) —,while L in (20) is defined in Appendix II.

A solution compatible with both sets and with equation (21) is shown in figure 3: the
two versions, right- and left, of the same electromagnetic field arise at each side of the

12
boundary. The t-forward wave E, B, kˆ +   R
forms a right-hand triad, and subsequently

the t-backward wave must be the left-hand triad5, E, B, kˆ    L

 t (back ) t ( forw) 
E L B E R B (22)

At the t-boundary itself,

E  E t 0

B   B  t 0
kˆ +   kˆ 
t 0

― t  0 ‖ represents here the ―time-zero‖, or temporal boundary, which, as this result

suggests, cannot be considered a mere, arbitrary selection of our timer. On the contrary,
it is a highly extraordinary circumstance where the field would perform an extraordinary
change of handedness, never so far observed (see section 6).

Note that the handedness symmetry between the equations (19) and (20), is fully passed
on to the solutions (22), with no need for further conditions or considerations about how
we imagine the time-backward field might be. As shown in figure 3, the effect of a
temporal boundary is to mirror the field, instead of mirroring the charge motion6.

Figure 3: The two, t-forward and t-backward, waves, emerge at both sides of a t-boundary with
the opposite handedness. The t-forward wave is the conventional electromagnetic wave, i.e., the
right-hand triad EBk. The symbols R and L stand for right and left, and represent the advance of
a right- or left- screw.

It is interesting to recall that the handedness convention is arbitrary, which always

allows for a definition of the field in two alternative forms, right or left, depending on

5
With time forwards, E rotates towards B according to a right-hand screw which advances to the k̂ + -
direction. With time backwards, E rotates towards B according to a left- handed screw, which points to
the k̂  -direction.
6
This different result comes from the essentially different approach: the conventional one adopts the
charge motion as ―the prime reality‖, and the field is subordinated to it [14]. For this paper‘s, the charge
is not even involved: the prime reality is the field [15].

13
the convention used: for the t-forwards, for example, the use of the right-hand is
generalised, and hence the use of the E, B, kˆ  
R
triad, but it would also be possible
to have used the opposite, left handedness, with which the field would be equally
defined by the left -handed triadB, E,kˆ  . Identically, for the t-backwards, the

L

above-mentioned left-handed triad E, B, k

ˆ  , or alternatively, the B, E,kˆ 
 
L R
right-handed one, equally represent the t-backward field, by virtue of the simple
algebraic relations between both handedness types (see Appendix II). This allows us to
rewrite (22) in any of the following forms:

Temporal Boundary
 t (backwards ) t ( forwards ) 
E L B =  B L E = B R E E R B =  B R E = B L E
(23.a) (23.b)

As expected, the double symmetry of the equations is also fulfilled by the solution:

Temporal Boundary

 t (backwards) t (forwards) 
F-symmetry

E L B = B R E B L E  E R B

H-symmetry
where H stands for Handedness symmetry (right-left) R  L , and F stands for Field
symmetry, E  B . We have reached this by performing the first operation H.
Obviously, the same result should be obtained after the second one F, since they are
equivalent.

3.2.2 Exchange of fields

Equally, the t-forward set (16) turns into the t-backward set (17), by performing an
exchange of fields E  B . Both sets are formally identical, with E and B renamed.
As it is obvious that mathematics do not penalize the name-exchange, the plane-wave t-
backward set must be formally identical to (19), with E and B renamed:

14
 t (forwards)   t (backwards)
kˆ   E  = 0 kˆ  B = 0
 

kˆ  B = 0
  kˆ   E  = 0
kˆ  R E = B  (24.a) kˆ  B = E
 R   (24.b)
k̂  R B  =  E  kˆ  R E  =  B 

E+ B
where 1 and 1
B+ E

Again, subscripts + and – highlight t-directions, (despite being clear in the top line of
the equations). Note that all the cross products are now right-handed, since the
handedness, in this case, has not been modified. For the t-forward set (24.a), the field is

E, B, kˆ  triad, and so the t-backward field must be the (also

the (right-hand) 
R

right-hand) B, E,k

ˆ  triad (with vectors E and B exchanged). As expected, this

R
result was already obtained —equation (23) —, and shown in figure 3. The t-backward
field, B, E,kˆ  is the dual of the t-forward field, E, B, kˆ 

R

R
. The term ―dual‖
is not arbitrary: it is borrowed from tensor analysis, with the same meaning7, (see
section 4). Hence, the ―electromagnetic‖ field going backwards in time is actually a
―magnetoelectric‖ field, or let us call it, the dual electromagnetic field, which is
composed of the dual electric (the former B) and the dual magnetic (the former E). They
have been indicated with apostrophes (figure 4):

Temporal Boundary

 t (backwards ) t ( forwards ) 
E  L B = B  R E = E'  R B' E  R B = B  L E = E'  L B'
(25.a) (25.b)

At the t-boundary, they fulfil:

E   B' t 0

B   E' t 0

kˆ +   kˆ 
t 0

7
The dual tensor is the resulting field tensor, after the field exchange, (see Appendix III).

15
With this notation, the symmetry of the solution is overwhelming: the electromagnetic
field could be defined, either by the conventional wave, or by its dual, with the opposite
handedness. At the temporal boundary, the field just transforms into its dual (I will
recall this result in section 4):

Temporal Boundary
t  t 
H*F
E L B E R B
H+F
E'  R B' E'  L B'

It is paradoxical that, contrary to what happens with space inversion, Maxwell‘s

equations (and their solutions) are not time-invariant because they do distinguish
between t+ and t-. And yet, they are double-symmetric (anti-symmetric, actually). The
apparent asymmetry of radiation we observe in Nature, surprisingly seems to come from
a double symmetry, if only we unfold the time dimension, in the two, backward and
forward, directions.

Figure 4: Field time-symmetry, in terms of the dual fields, which have been identified
with apostrophes. Again, the symbols R and L stand for right and left, and represent the
advance of a right- or left- screw.

3.3 Comparison of space and time inversions

If we compare how Maxwell‘s equations transform in spatial and time inversions, we

find an essential difference between the two cases: from the final equations in section 2,
it is clear that Maxwell‘s equations do not change in a spatial inversion, while they do in
a time inversion, if we just have a look at equations sets (16) and (17) in section 3. From
the point of view of the electromagnetic field, one subspace is identical to the other. The
directions of time, though, are not. Mathematically, the space inversion maintains the
handedness, whichever it is, (even H-symmetry), while the time inversion does not (odd
H-symmetry). Physically, this means that the locations of sources and sinks are
exchangeable, but the transmutation of sources into sinks and vice versa, would require
a mirror image of the conventional electromagnetic wave [15]. Such a behaviour has not

16
been observed so far in Nature, which may indicate, either that the electromagnetic field
is not able to revert in time, (contrary to what is usually assumed), or, if it is so, that
time boundaries are hardly accessible (see section 6).

Spatial Boundary

s  s 
E R B E R B
H-even

Temporal Boundary
t  t 
E L B E R B
H-odd

4. Four-dimensional formulation
The four-dimensional formulation has been also used to justify the invariance of
classical electromagnetism [17]. Indeed, if one starts from the same hypothesis (the
identification of the field time reversal with the charge motion reversal), obviously will
get the same result, expectedly invariant, with any of both formulations, classical or
four-dimensional, since they are equivalent. The way in which the conventional time
inversion ( B 
 B ) can be translated into the latter formulation:
T

F 
T
 F 
(Conventional)

F  
T
 F

i.e., the Maxwell-Faraday tensor transforms into its contravariant form (see Appendix
1
III). This makes the quantity F F  ,—which is an invariant of
2
electromagnetism—, to stand in the time inversion, since its value is given by (natural
units):
1
F F   B 2  E 2
2
and obviously, it does not change under a sign flip, since the fields are squared. This is
the way in which the four-dimensional formulation seems to confirm the premise of
electromagnetic invariance under a time inversion. The difference is found, though, if
we start from a different premise: if the time inversion is taken, not as the sign-flip

17
B 
T
 B , but as the one proposed here, E 
T
 B , the correct tensor
transformation would be:

F 
T
 G 
(this proposal)

F  
T
 G

i.e., the Maxwell-Faraday tensor turns into its dual, —a result which is a natural
extension of the one already obtained in the previous section 3.2.2. With this, the
1 1
quantity F F  does not hold any longer: Its value, F F   B 2  E 2 , turns
2 2
into:
1 1
F F  
T
 G  G  E 2  B 2
2 2
i.e., its sign flips8, (just applying the exchange of fields to the standard value). If this is
1
correct, the variability of the quantity F F  , otherwise invariant, may be another
2
symptom that the electromagnetic field is indeed sensitive to time inversions, contrary
to the common opinion.

5. Discussion of the solution

As put forward in section 3, Maxwell‘s equations with time backwards (15) are
normally solved by reassigning the negative sign coming from time backwards to one of
the fields. We find in literature two options [16]:

1. The textbook approach, which reassigns time‘s minus to the B field (while E does not
change):

B 
T
 B
(E 
T
 E)

2. The alternative, Feynman‘s approach, which reassigns time‘s minus sign to the E
field (while B does not change):

E 
T
 E
(B 
T
 B)

8
Under the conventional time inversion, this quantity behaves as a proper scalar. Under this paper‘s, it
behaves as a pseudo scalar. This sign flip will also affect the electromagnetic Lagrangian with time
backwards.

18
The textbook approach considers the effect that a charge moving backwards would
produce on the magnetic field. Feynman‘s, instead, considers the cause, i.e., the fact
that, for a charge to reverse its motion, it would require a reversed electric field.

The time inversion proposed in this paper does not mean any sign flip in the fields, but
an exchange of roles between the electric and magnetic fields themselves9:

E 
T
B
(in natural units)
B 
E T

It is possible to justify this result, in comparison with the abovementioned approaches.

Basically, the time backwards alters the normal sequence of events, and the cause-effect
sequence is also inverted. With time forwards, the electric field (the cause) produces the
charge motion, which, at the same time, produces the magnetic field (the effect). With
time backwards the opposite is expected: the magnetic field (the effect) ―counter-
produces‖ the charge motion reversal (remember, with time backwards), which, at the
same time ―counter-produces‖ the electric field (the cause). With this new viewpoint,
the exchange of roles between E and B would somehow be justified because, with time
backwards, the effect (B) is prior to the cause (E), and responsible for the charge
motion10.

t  forwards t  backwards
cause  effect cause  effect
E  motion charge  B E  motion charge  B

6. But what is a time border?

For the conventional time-inversion, a time border would just be when the magnetic
field suffers a phase shift B  B , crossing its zero-value (something relatively
common, so much that anyone could force that event to happen any time) For this
paper‘s, time borders are not such a common event, and of course not any arbitrary
choice of our timer. Instead, they would be extraordinary time points where the
electromagnetic field suffers an extraordinary exchange of roles, implying at the border
itself the degeneracy of the electric and magnetic fields E=B . Indeed, they would be
exceptional events, able to provide an absolute reference for the dimension of time, if
measured or defined in relation to them.

9
The double symmetry should not be mistaken with the electric-magnetic duality, by means of which
E  B and B  E , leaving Maxwell‘s equations invariant [18]. Being so, the two directions of time
are indistinguishable, something which fulminates any hope of hosting time singularities, such as, for
instance, the origin of the time arrow. The double symmetry precisely reveals how Maxwell‘s equations
can discriminate the two directions of time, allowing that possibility.
10
This would mean that the electric and magnetic fields would be the temporal mirror of each other.

19
Such time points, in my view, would be more in line with true time borders already
predicted in Cosmology, like the Big Bang or the absolute zero temperature. These two
points are good time-border candidates, not only for the obvious cosmological reason of
End and Beginning of Time they represent, but also for some phenomenology that has
been predicted for them.

6.1 The absolute zero-point temperature

Regarding the temperature of absolute zero (0 K), precisely its impossibility to be

achieved (according to the Third Law) suggests the same impossibility of going over the
limit of time backwards. In the vicinity of that point, time just slows down, as we see,
for instance, in the case of vital processes as temperature decreases. But there is a clue
which may support this idea in a more concrete way. Wu‘s famous experiment, where
parity breaks down, is performed just above (very near) the temperature of absolute zero
[19]. That could be precisely the key factor: near absolute zero the fields unexpectedly
degenerate, and the applied B starts to behave like an electric field, provoking the exit
of electrons in its opposite direction, just as an electric field would do (B‘s behaviour,
though, would not be perfectly electric, but a degenerate magnetic-electric mixture).

It is remarkable that, in that famous experiment where parity breaks down, left and right
start to become distinguishable, at the same time that the electric and magnetic fields
start to degenerate (and vice versa: far from the border, when E and B are
distinguishable, left and right degenerate). This indicates an intimate, balanced
relationship between space and time, so far unexplained.

6.2 The Big Bang

The well-known ―problem of the missing monopoles‖, which arises from the prediction
of the GUTs11 in relation to an abundance of magnetic monopoles in the Big Bang,
however, does not match with the fact of their inexistence now [18,20-21]. Despite
being partially solved by the hypothesis of inflation, this problem could be also
explained as a consequence of the field degeneracy at a time border.

Although the study of the charge behaviour is beyond the scope of this paper, note that
the degeneracy of the fields at a time border implies the degeneracy of the monopoles,
too. For the sake of economy of resources and symmetry, it can be deduced that two
different kinds of monopoles (electric and magnetic) would be redundant for a single
kind of field —the degenerated one— at the time border. The apparent distinction
between the two kinds of monopoles would come from the field duality, electric-
magnetic, which rules on each side. The distinction, therefore, would be rhetorical: there
would not be ―electric monopoles‖ or ―magnetic monopoles‖, but just ―monopoles‖,
driven by electric and/or magnetic fields, depending on which side of the boundary are
being observed from. At the time border itself, they would not be distinguishable, —just
like the fields—. Thus, it would not be surprising a massive presence of ―magnetic‖
monopoles at the Big Bang, (as well as ―electric‖): If the hypothesis in this paper is
correct, they might be all the same at the time border, and their ―mysterious‖

11
Grand Unified Theory.

20
disappearance afterwards would be the consequence of the ceasing of the field
degeneracy, which makes them appear like ―electric‖ monopoles from this side, quite
far from the time border. The asymmetry of matter, (represented by the mentioned
asymmetry of monopoles) would therefore be linked to the asymmetry of radiation,
being both only apparent, consequence of the partial viewpoint from only one side of
the time border.

6.3 Remarks on energy

The solution proposed in this paper suggests two waves (one at each side of the time
border) which only differ in the handedness. If energy conservation is not a matter of
handedness, the left-handed wave will fulfil energy conservation for time backwards, as
much —or as less— as the right-handed one does, for time forwards. This, again,
suggests the exceptionality of a time border, as a singular point where energy flips its
sign (however, not so singular as to violate energy conservation, if the energy in both
time directions keeps constant, see Appendix IV for further comments).

7. Conclusions
This paper deals with the problem of space and time inversions in classical
electromagnetism. The space inversion in this paper is not the conventional right-left
parity, but a generalization, in the sense that the inversion surface is not a plane, but a
general, closed surface. The surface generalization is precisely motivated by our aim to
explore the behaviour of classical electromagnetism in time, conventionally considered
as time-reversible, due to a questionable mechanicistic view of the electromagnetic
field.

Here it has been shown that Maxwell‘s equations in free space are space-invariant, but,
contrary to the general opinion, they are not time-invariant, once released the hypothesis
of subordination of the field to matter. Still, they show an overwhelming double
symmetry, handedness (H-symmetry) and fields (F-symmetry). This symmetry pattern
that the field shows in time suggests that the turning points of time (here called ―time
borders‖) are not common nor arbitrary, but exceptional time points where the
electromagnetic field performs an extraordinary exchange of roles, becoming
degenerated at the border itself. This degeneracy has consequences on the number of
monopoles required for the symmetry of classical electromagnetism, since it is
redundant to define two different kinds of monopoles for a single kind of field (the
degenerated one existing at the border). The asymmetry of matter and the asymmetry of
radiation are both linked, being only apparent, consequence of the partial viewpoint
from one side of the time border.

The behaviour of matter in time has been left unsolved in this paper. Not as obvious as a
simple backwards motion, it would imply the redefinition of monopoles under the new
field, and will have to be inferred from the time-symmetry demanded by the field. It
could be a nice topic for the future.

21
Appendix I: Space reversibility of Ampère-Maxwell’s equation, and the
continuity equation

If we consider the notation I and O introduced in section 3, Ampère-Maxwell‘s equation

can be written in terms of subspace I, as:

1   E  dS
  0  j  dS I
I
Bd = S
C I
c2 t S

In terms of subspace O, it could be written as:

1   E  dS
  0  j  dSO
O
Bd = S
C O
c2 t S

Obviously, both expressions are identical, by virtue of equations (3) and (4).

The same derivation can be done for the continuity equation. Observer I would write it
as:
dqI

S
j  dS I  
dt

Taking the derivative of equation of equation (2), dqI   dqO , the same continuity
equation also stands for observer O:

dqO
 S
j  dSO  
dt
This means that none of both points of view is preferred or more special than the other:
there is a total symmetry between the inside and outside for Maxwell‘s equations.

Appendix II: Handedness in the vector product

A curious property of the cross product is that it is anti-commutative for same
handedness,

A  R B =  B  R A,
A L B =  B L A

22
and commutative for opposite handedness:

A  R B = B  L A,
A L B = B R A

The right cross product  R is the conventional cross product  , based on the right
handed system, {ux, uy, uz}. The left product L is based on the mirror image of the
right handed system, {ux, uy, uz}:

ux uy uz
A  L B  ax ay az 
bx bx bz 

where uz uz , and {ax, ay, az} and {bx, by, bz} are respectively the
components of A and B in the left-handed system.

Appendix III: Notation for the four dimension formulation

The Maxwell-Faraday tensor F is given by [22]:

 0  cBz  cBy  Ex 
 
  cBz 0  cBx  Ey 
F 
 cBy  cBx 0  Ez 
 
 E  Ey  Ez 0 
 x

And the contravariant form, F  :

 0  cBz  cBy  Ex 
 
  cBz 0  cBx  Ey 
F  
 cBy  cBx 0  Ez 
 
 E  Ey  Ez 0 
 x

Note that both are related by one field‘s sign flip, which represents the conventional
time inversion.

23
The dual field tensor G  is given by [22]:

 0  Ez  Ey  cBx 
 
  Ez 0  Ex  cBy 
G  
 Ey  Ex 0  cBz 
 
  cB  cBy  cBz 0 
 x

The transformation between F and G  is precisely the field transformation for the
time inversion proposed in this paper12. The covariant components, G  :

 0  Ez  Ey  cBx 
 
  Ez 0  Ex  cBy 
G  
 Ey  Ex 0  cBz 
 
  cB  cBy  cBz 0 
 x

Appendix IV: Further comments on energy

At first glance, the statement that classical electromagnetism is not time-invariant might
seem to contradict the energy conservation law, since the latter precisely means the
invariance of natural laws in time. The time-invariance implied by energy conservation,
though, is equivalent to the homogeneity of time, (instant by instant, all are identical),
but not to its isotropy (direction by direction, may be not).

The dichotomy between homogeneity and anisotropy of Time appears, at least, in one
instant of Time: the Big Bang (when dealing with the Time‘s arrow, sooner or later you
bump inevitably into its origin). That is an accepted singularity, where Time is not
homogeneous. Should, then, the Big Bang be excluded from energy conservation? How
to combine the Time‘s homogeneity for positive time, with the Time‘s singularity at the
Big Bang?

This question of how the Big Bang fits into the energy conservation law is far beyond
the scope of this paper, but at least we need to justify the idea that both the homogeneity
and the anisotropy of Time are not incompatible (or say, the non-time invariance of
Electromagnetism proposed in this paper does not contradict the energy conservation
law).

In section 3.2.1, two identical waves (except for the handedness) arose at both sides of
the time border, hence carrying opposite energies in each time direction. This result is
compatible with the energy conservation principle, if ―refined‖ with the specification of
a value (zero) for the net energy of the Universe, in the same way that it was done with

12
This paper‘s time-inversion, in SI units, E 
T
 cB

24
the charge of the Universe in section 2. Thus, the energy conservation law (the energy
of the Universe is constant), would become the energy neutrality of the Universe, (such
a constant should be zero). In agreement with this, a possible energy model for a time
singularity is displayed versus time in figure 5. At the time-border, energy exits (source-
type border, figure 5.a), —or enters, sink-type border, figure 5.b—, towards —or
from— the two directions of Time. The energy is constant for each direction of time
independently, ( E , for any instant after the Time‘s border, and E for any instant
before it, hence the homogeneity of Time observed for positive time), and, in virtue of
energy neutrality, both cancel each other out:

E  E  0  E   E
This means that energy neutrality requires that the energy travelling forwards in time
must be opposite (different) to that travelling backwards (hence the anisotropy of Time
at the time border).

Note that time borders, in this paper, are presented as exceptional, singular points,
which, however, are compatible with natural laws (Time is singular, but natural laws are
not). This is a significant difference with the conventional view of the Big Bang, which
is commonly taken as the singular instant where Time starts (there was nothing before,
neither Time nor natural laws, i.e., a moment of (absolute) ―creation‖, as we could call
it). The solution in this paper suggests a different time border concept, not as a ―moment
of creation‖ (or ―destruction‖), but as a moment of transformation, precisely a
consequence of the symmetry of natural laws in time.

Obviously, this is the simplest model: one instant‘s singularity and a perfectly
homogeneous time afterwards. But the admission of energy neutrality for one instant
makes it possible for other instants too. Figure 5.c, for instance, shows a not so neat
time border.

Figure 5: Energy versus time, (simplified model, one-instant singularity). a) Source-type time
singularity. b) Sink-type singularity. c) Broad singularity (in grey, the singularity region).

25
 References
[1] Wigner, E.P. [1965] Violations of Symmetry in Physics, Scientific American,
213: 28-36.
[2] Loschmidt, J. [1876]: Math Naturewiss, II-73, 128.
[3] Wheeler, J. A., Feynman, R.P. [1945] Interaction with the Absorber as a
Mechanism of Radiation, Rev. of Modern Physics, 17: 157-181.
[4] Price, H [1996], Time’s Arrow and Archimedes’ Point: New Directions for the
Physics of Time. New York: Oxford University Press.
[5] Zeh, H.D. [1999]: The Physical Basis of the Direction of Time, 3rd Ed. Berlin:
Springer-Verlag.
[6] North, J. [2003] Understanding the Time Asymmetry of Radiation, Philosophy
of Science, 70(5): 1086-1097. Retrieved from
http://www.jstor.org/stable/10.1086/377391
[7] Jackson, J.D. [1962], Classical Electrodynamics, New York: John Wiley &
Sons.
[8] Schwinger, J., Deraad, L.L., Milton, K.A. [1998], Classical Electrodynamics,
Reading, MA: West View Press.
[9] Albert, D.Z. [2000] Time and chance, Cambridge (MA), Harvard University
Press.
[10] Frisch M., [2000] (Dis-)Solving the Puzzle of the Arrow of Radiation, Brit.
J.Phil. Sci. 51: 381-410.
[11] Orito, S. Yoshimura, M. [1985]. Can the Universe be Charged?. Phys. Rev. Lett.
54 (22): 2457–2460.
[12] Masso, E. Rota, F. [2002]. Primordial helium production in a charged universe.
Phys. Lett. B 545 (3-4): 221–225.
[13] Sakurai, J.J., Napolitano, J. [1994] Modern Quantum Mechanics, p.284, 2nd Ed.,
San Francisco, CA: Addison-Wesley.
[14] Wigner, E.P. [1932] Nachr. Ges. Wiss. Götingen p.546.
[15] Masot-Conde, F. [2014] Radiation Arrow and Missing Magnetic Monopoles,
retrieved from https://zenodo.org/record/12999#.VQLEWI6G_eI
[16] Arntzenius F., Greaves, H., [2007] Time reversal in classical electromagnetism.
Retrieved from http://philsci-archive.pitt.edu/3280/
[17] Malament, D. B. [2004] On the time reversal invariance of classical
electromagnetic theory. Studies in History & Phil. Modern Physics, 35, 295-315.
[18] Rajantie, A. [2012] Introduction to Magnetic Monopoles. Cornell University
Library. Retrieved from http://arxiv.org/abs/1204.3077
[19] Wu, C. S.; Ambler, E.; Hayward, R. W.; Hoppes, D. D.; Hudson, R. P.
[1957]. Experimental Test of Parity Conservation in Beta Decay. Physical
Review 105 (4): 1413–1415.
[20] Berstein, J. [1989] The Tenth Dimension, New York: Mac Graw Hill.
[21] Salam, A. [1966] Magnetic monopole and two photon theories of C-violation,
Phys. Lett., vol. 22, (5), p. 683-4.
[22] R. Fitzpatrick, [2008], Maxwell Equations and the Principles of
Electromagnetism, Hingham, MA: Infinity Science Press.

26