Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

(Pre-) causality and the Lorentz Transformations

Enrico Menotti*
* via del Castelliere 23, 34076 Romans d’Isonzo (GO), Italy, enrico.menotti@libero.it

ABSTRACT

The Lorentz transformations may be derived based on the Principle of Relativity and a few other plausible physical assumptions,
without introducing the Principle of the Constancy of the Velocity of Light. Such a procedure is not new in the literature—it
appeared first in the pioneering paper by W. von Ignatowsky published on Phys. Z. 11, 972 (1910); yet, all existing derivations
of this kind introduce a (pre-) causality (or equivalent) condition in order to constrain the transformations of coordinates to
assume the form of the Lorentz transformations (or their Galilean limit). We show how such a condition may be dropped: as a
consequence, the Principle of Relativity in itself implies any causal relationship to be conserved under a change of reference
frame.

Introduction
In his famous 1905 paper,1–3 Einstein developed Special Relativity as a principle theory: by assuming a few general properties
to be valid in nature, he was able to deduce the relationship between the points of view of different observers in relative motion
with respect to each other. Indeed, he worked out the transformations of coordinates (known as the Lorentz transformations) on
the basis of two fundamental principles: the Principle of Relativity and the Principle of the Constancy of the Velocity of Light.
In the investigation for a minimal set of principles, many derivations of the Lorentz transformations have been developed which
do not make use of the Principle of the Constancy of the Velocity of Light (an early attempt may be found in Ref.4–8 ; later
independent re-discoveries are presented in Refs.9–11 ; recent discussions are those of Refs.12–15 ); such a procedure is possible,
but leads invariably to a set of transformations which depend upon an unknown (real) parameter. As far as we know, all these
derivations allow for a negative value of the parameter, and then rule out the corresponding transformations on the basis of
further (physically reasonable) assumptions such as (pre-) causality (some authors—see, e.g., Ref.11 —call this assumption
simply “causality;” others—e.g. Refs.12, 14 —introduce the word “precausality.” In order to include both definitions, we use the
terminology “(pre-) causality”). As stated in Ref.11 , this is the requirement that “the nature of a possible causal relationship is
not changed under inertial transformations;” in other words, “if two events take place at the same point in space with respect to
a given observer, their time order must be the same for all observers.”9, 10, 12, 14
In our opinion, (pre-) causality is needed due to insufficient consideration of the group properties imposed by the Principle
of Relativity. The aim of this paper is to show how non-causal transformations (corresponding to negative values of the
above-mentioned parameter) are already excluded by these properties.
An immediate clarification is worth. As will be stressed later, we are dealing with a mathematical description of a
physical setup, not with a mere mathematical problem. If one only considers group properties in an abstract framework,
our statement is wrong. But we are considering transformations between physical reference frames, not between abstract
(two-dimensional) spaces. This will lead to the preliminary condition that the set of transformations be parametrised by v,
the relative velocity between the two reference frames under consideration. Once group properties are imposed on this set,
non-causal transformations are ruled out.

Physical set-up
In the spirit of the original work by Einstein, we assume some (point-like) body to be the origin of a system of coordinates;
(space) distances are defined by means of “rigid” rods and (time) intervals by means of “ideal” clocks (suitably synchronised
according to some convention). This point has been widely debated in the literature.16 For the sake of simplicity we will only
consider a single spatial coordinate, so that we have to find a relationship between couples of (space-temporal) coordinates.
Be the matter settled in this way, let us consider two systems of coordinates (x,t), (x0 ,t 0 ) such that their (collinear) spatial
axes have the same orientation and their origins coincide, i.e. when a clock (at rest) at x = 0 reads t = 0 the (hypothetical) body
marking x0 = 0 is located in x = 0 and a clock (at rest) at x0 = 0 reads t 0 = 0—in short, x,t = 0 implies x0 ,t 0 = 0. We assume the
second system to be in uniform relative motion with respect to the first: let v be the velocity of the body marking x0 = 0 as
measured in the first system, v is constant (in t). Such a change of reference frame is called a boost.

1
This is the right point to state our first mathematical condition, which is physically unavoidable, and (mathematically) plays
the role of (pre-) causality. We need to use an appropriate terminology to distinguish clearly between physical reality and its
mathematical description: we will use the word “phenomenon” to indicate a physical event, and refer to the corresponding
mathematical entity as its “coordinates.” The term “event” is a bit ambiguous (in common language it refers to something
which happens in the physical world, while in the context of Special Relativity it usually denotes an element in a mathematical
space), and as such will be avoided.
A system of coordinates is a mathematical description for phenomena (points in space at instants in time) such that, given
an phenomenon, a couple of coordinates (its space distance and time delay or advance with respect to the origin) is determined.
Note that we are deliberately considering cartesian coordinates, with a single spatial component; i.e., the space coordinate for a
phenomenon is simply its distance from the origin, and the time coordinate is the interval between the phenomenon and the
origin (precisely, the phenomenon at the spatial origin which marks the initial time instant). This choice for the coordinates
of a phenomenon implies that such coordinates are unique, as far as distances and intervals are well defined. In other words,
given a phenomenon, its coordinates (in a system) are uniquely determined. The converse is true as well: by assumption, we
are considering a single spatial coordinate; in other words, we assume a phenomenon to be uniquely determined by its two
coordinates. Shortly, given a system of coordinates, there is a one-to-one correspondence between phenomena and coordinates.
Now, we have two systems of coordinates. In order to fix ideas, consider the first “at rest.” In other words, let its origin
be well defined (it is a certain phenomenon). Distances and intervals in this system are also defined by some operational
rule. As discussed, phenomena are in a one-to-one correspondence with coordinates in this system. Let v, the velocity of
the second system (as measured in the first), be fixed. Then we have the law of motion for the spatial origin of this “moving”
system: to each time coordinate there corresponds a (unique) spatial coordinate for this origin, which (uniquely) identifies
a phenomenon; in other words, the spatial origin of the moving system spans phenomena in such a way that to each time
coordinate (in the system at rest) there corresponds a unique phenomenon. This means that the physical setup is completely
defined; in other words, the origin of the moving system is well defined (given a time coordinate in the system at rest, it is a
certain phenomenon). Together with the operational rule which defines distances and intervals in the moving system, this puts
phenomena in a one-to-one correspondence with coordinates in this system.
At this point, we have the following situation: two systems of coordinates are well defined; i.e., phenomena are in a
one-to-one correspondence with (space-temporal) coordinates in both systems. Then, to each couple of coordinates in the
system at rest there corresponds a unique couple of coordinates in the moving system (determined by the corresponding
phenomenon), so that the transformation of coordinates from the first to the second system is uniquely determined. In short,
once v be fixed, the transformation is identified, i.e. the set of transformations is parametrised by v.
Clearly this discussion does not take into account the Principle of Relativity, as stated later. In the view of such a principle,
we cannot consider any system of coordinates as being “at (absolute) rest.” However, this does not invalidate our conclusion:
once we select a certain system as our reference “at rest,” we do characterise phenomena by their coordinates in this system.
In our opinion, this procedure allows the above considerations, so the conclusion that v be a good parameter is valid and the
following dissertation, which is based on this conclusion, is correct.

Homogeneity of space and time and isotropy of space


We will show that the transformations of coordinates assume the form of the Lorentz transformations (with the actual value
of the invariant speed being unknown and possibly infinite, in which case they reduce to Galileo’s transformations) when
constrained by just three assumptions. The first is the homogeneity of space and time; this is a natural assumption (at least at a
macroscopic level) and is sufficient to ensure linearity of the transformations. A simple proof of linearity may be found in
Ref.11 (other proofs are available in the literature; see e.g. Refs.9, 10, 17 ). We will not report such a proof here; it is enough to
write down formulae (10) as found in Ref.11 :
(
x0 = γ (v) (x − vt) (1a)
t 0 = γ (v) (λ (v)t − µ (v) x) , (1b)

where γ, λ and µ are (real) coefficients depending on the parameter v. In formula (1a) use has been made of the definition of v.
Note that, due to the Principle of Relativity as stated later, the functions γ, λ and µ are the same for all couples of reference
frames. The mathematical expression of the Principle of Relativity is precisely that the transformations of coordinates only
depend on v and not on the choice for the unprimed reference frame; all other consequences may be derived from this one.
The second assumption is the isotropy of space, which is also natural; in our (spatially) one-dimensional set-up it amounts
to say that the two possible orientations of the space axis are equivalent.9–11, 14 Again, we will directly report the mathematical
consequences of isotropy; these are expressed by equations (15) in Ref.11 , which are the following symmetry conditions on the

2/7
coefficients of formulae (1):

γ (−v) = γ (v) (2a)


λ (−v) = λ (v) (2b)
µ (−v) = −µ (v) (2c)

(note that condition (2c) implies µ (0) = 0).

Group structure
The third assumption needed in order to obtain the Lorentz transformations is the Principle of Relativity; in short, it states
that there is no absolute motion, i.e. motion is only defined relative to some body (taken as a reference).1 This is reflected by
an (algebraic) group structure in the set of all possible coordinate transformations,9–12, 14, 15 which means that the following
conditions are met.
i) The set of all possible boosts is equipped with a (binary) operation (namely, composition of boosts) which leads from any
couple of boosts to a novel (possible) boost;
ii) an identical element exists: any boost, when composed with this element, remains unchanged;
iii) each boost has its inverse, such that when the two are composed the result is the identical element.
Physically, condition i) amounts to require that the composition of any two (possible) boosts be a (possible) boost, condition ii)
is the requirement that when v = 0 the boost leave the reference frame unaltered and condition iii) means that the inverse of any
(possible) coordinate transformation must be a (possible) boost.

Reciprocity Principle
Let us call S, S0 the systems of coordinates (x,t), (x0 ,t 0 ), respectively. The (direct) motion of S0 with respect to S is uniform,
with a velocity v. What is the (reciprocal) motion of S with respect to S0 ?
The Lorentz transformations are quite easily found if it is assumed that this reciprocal motion is uniform as well, with a
velocity −v (Reciprocity Principle). Many derivations assume this principle to be a consequence of the Principle of Relativity
(alone) (see, e.g., Refs.5, 13 and the references in Ref.10 ); as discussed in Ref.10 , this is not true: the Principle of Relativity
merely implies the relation between direct and reciprocal motion to be the same for all couples of reference frames—indeed,
theories have been proposed in which clocks are synchronised in such a way to allow for space anisotropy: they are examples
of theories (based on the Principle of Relativity) in which the Reciprocity Principle does not hold. See the discussion in Ref.14
and related references.
Actually, the Reciprocity Principle is a consequence of our three assumptions—homogeneity of space and time, isotropy of
space and the Principle of Relativity—together with two continuity hypotheses. In order to specify the latter, let us call Γ the
set of allowed values for v; due to the Principle of Relativity, Γ is the same for all couples of reference frames. The needed
continuity hypotheses are:
a) Γ is a (real) interval;
b) γ, λ , µ are continuous on Γ.
(to be precise, the continuity of γ and µ is not needed in the proof of the Reciprocity Principle; the continuity of γ is used
later—it is the fundamental assumption which rules out non-causal transformations—while that of µ may be dropped). A proof
of the Reciprocity Principle may be found in Ref.10 . Let us briefly review this proof.
The first thing to do is to express the velocity u of S with respect to S0 in terms of v and of the yet unknown function λ . This
is readily done from formulae (1) by imposing x = 0 and eliminating t (u is the velocity of the body marking x = 0 as measured
in S0 ); a straightforward calculation yields u as an (unknown) function ϕ of v alone, so u is constant (in t 0 ), which means that S
is in uniform relative motion with respect to S0 . Shortly, u = ϕ (v), where, explicitly, ϕ (v) = −v/λ (v).
At this point ϕ is worked out in terms of v: this provides the requested relation between v and u and the form of λ . The
procedure relies on some mathematical properties of ϕ, which are deduced from physical considerations; namely, it is found
that ϕ is a bijection of Γ in itself, whose inverse is ϕ itself. But this is not enough—three more properties of ϕ are required. The
first is deduced from hypothesis b): since λ is continuous on Γ, such is ϕ as well. The second comes from condition (2b), which
1 This form for the Principle of Relativity is found in Ref.5 .

3/7
implies the antisymmetry condition ϕ (−v) = −ϕ (v). The third requires a short digression on condition ii)—by substituting
v = 0 in formulae (1) and requiring that they reduce to identical transformations (in this situation S0 coincides with S), we
readily find the following equivalent conditions:

γ (0) = 1, (3a)
λ (0) = 1. (3b)

The second of these implies, in particular, that λ (0) be positive; since λ is continuous, we may deduce it to be positive in some
neighbourhood of v = 0. As a consequence, in that neighbourhood ϕ (v) > 0 if v < 0 and ϕ (v) < 0 if v > 0.
There is a nice and simple theorem in mathematical analysis which states that a continuous bijection defined over an interval
is strictly monotonic: it follows that ϕ is (strictly) decreasing. Based on this, the antisymmetry of ϕ and the property ϕ −1 = ϕ,
it is easily shown that ϕ (v) = −v, which proves the Reciprocity Principle.
The result just stated allows immediately to find λ (v) = 1, which, substituted into formulae (1), yields:
(
x0 = γ (v) (x − vt) (4a)
t 0 = γ (v) (t − µ (v) x) . (4b)

Imposing the group structure


We have seen that the Principle of Relativity amounts to impose conditions i), ii) and iii) on formulae (1). Part of this job has
already been done: on our way to the Reciprocity Principle we have made use of condition ii) in order to obtain conditions (3),
the second of which has been employed and is already satisfied by formulae (4). Now we exploit conditions i) and iii) to further
constrain formulae (4); in the next section we will conclude by considering condition (3a).
Before proceeding, it is convenient to write formulae (4) in a compact form by means of a matrix:
 0   
x 1 −v x
= γ (v) . (5)
t0 −µ (v) 1 t

We start by condition iii): due to the Principle of Relativity, the boost from S0 to S—which is described by the inverse of the
coordinate transformation leading from S to S0 —must take the form (5) (with x ↔ x0 , t ↔ t 0 and v → u). The first form is the
inverse of formula (5):
     0
x 1 1 v x
= ; (6)
t γ (v) (1 − vµ (v)) µ (v) 1 t0

the second form becomes, based on the Reciprocity Principle (u = −v) and conditions (2a, 2c),
     0
x 1 v x
= γ (v) . (7)
t µ (v) 1 t0

By comparison between (6) and (7), we obtain:


1
γ 2 (v) = . (8)
1 − vµ (v)

As a side remark, note that (8) implies the determinant of the transformation matrix in (5) to be 1.
Condition i) requires considering a couple of boosts: let S, S0 and S00 be three reference frames, with S0 in uniform relative
motion with respect to S at a velocity v1 and S00 in uniform relative motion with respect to S0 at a velocity v2 . Due to the Principle
of Relativity, both of the boosts from S to S0 and from S0 to S00 are described by formula (5), with the substitutions v → v1
and {(x,t) → (x0 ,t 0 ) , (x0 ,t 0 ) → (x00 ,t 00 ) , v → v2 } respectively; by composing the resulting formulae we obtain the following
description for the transformation leading from S to S00 :
 00    
x 1 + v2 µ (v1 ) − (v1 + v2 ) x
= γ (v1 ) γ (v2 ) . (9)
t 00 − (µ (v1 ) + µ (v2 )) 1 + v1 µ (v2 ) t

The velocity V of S00 with respect to S is obtained from formula (9) by setting x00 = 0, and it is readily seen that it is constant (in
t). As a consequence, the transformation leading from S to S00 is a boost, so (9) must take the form (5) (with (x0 ,t 0 ) → (x00 ,t 00 )
and v → V ); in the latter, the diagonal elements of the matrix are equal to each other, so such must be the diagonal elements of

4/7
the matrix in (9) as well: this amounts to say that v2 µ (v1 ) = v1 µ (v2 ), i.e. µ (v) = kv, where k is a (real) constant. By the way,
note that this result immediately yields the formula for the composition of velocities:
v1 + v2
V= . (10)
1 + kv1 v2
Substitution of µ (v) into (8) yields:
1
γ 2 (v) = . (11)
1 − kv2
Now γ (v) is a real number, so γ 2 (v) is nonnegative; by imposing this condition on equation (11), we obtain kv2 < 1. Provided
satisfied (if k > 0, this sets a constrain on the boundaries of Γ), we can express γ from (11): let v ∈ Γ, we
this inequality is √
have γ (v) = ±1/ 1 − kv2 and, since Γ is an interval and γ is continuous, the sign is the same for any v. At this point all group
conditions have been exploited except (3a); this last condition will provide us with the information necessary in order to find
the function γ and constrain the sign of the unknown parameter k.
The first part (finding γ) is straightforward: condition (3a) implies, in particular, that γ (0) be positive; according to the
discussion at the very end of the previous paragraph, we may conclude that the (v-independent) sign in the expression of γ is
positive. Thus,
1
γ (v) = √ . (12)
1 − kv2
This formula satisfies condition (3a), so all group conditions have been used. What is left to do is to express the boundaries of
the interval Γ, which depend on k: we wish to discuss how this dependence vary in terms of the sign of k itself.
First, in the case k = 0 there is no constrain on Γ (it is the whole real axis), and formulae (4) reduce to:
(
x0 = x − vt (13a)
t 0 = t, (13b)

which are the classical Galileo’s transformations. √


Now we investigate the situation of k > 0. In this case, the inequality kv2 < 1 reads |v| < c, with the position c = 1/ k
(note that µ (v) is dimensionally the reciprocal of a velocity, so k is the reciprocal of a square velocity and c is a velocity). Then
Γ = ]−c, c[, k = 1/c2 and µ (v) = v/c2 , and formulae (4) become:
 0
x = γ (v) (x − vt) (14a)
 v 
t 0 = γ (v) t − x , (14b)
c2
with γ given by formula (12):
1
γ (v) = q . (15)
2
1 − vc2

These are the Lorentz transformations.

A novel approach to the case k < 0


We now come to our point of view on the situation of a negative parameter k. We will show how this ends up in a pathological
structure for the set of allowed transformations (namely, it reduces to the trivial group whose only element is the identical
transformation). This result follows from group structure, without the need for further assumptions. The fundamental idea may
be found in Ref.10 : there, the transformations corresponding to k < 0 are explicitly recognised not to form a group. This should
be enough to rule out these transformations. Conversely, this and all similar derivations which we have been able to locate in
the literature feel the need to consider the full group, and then introduce (pre-) causality (or equivalent conditions) in order for
this case to be excluded.
The first point to be clarified is the parametrisation of the set of transformations. We already discussed extensively this
topic—the set of transformations is parametrised by v. Here we stress that, to our knowledge, all Authors agree on this
parametrisation, at least at the outset, and possibly after having discussed linearity of the transformations. E.g., Ref.11 shows

5/7
first that the transformations are linear, then immediately introduces v as a parameter (see formulae (10)). The same is found in
Ref.10 (see formulae (8)). Yet, some Authors (e.g., Ref.10 ) switch to a different parameter when considering the case k < 0 (so
to have a full group). In our opinion, this is not allowed (unless the new parameter is in a one-to-one correspondence with v).
Now, Ref.10 provides a comprehensive account of the properties of our transformations (16). Right after their formulae (43),
these Authors write:
...transformations (42) are ordinary circular rotations. Since α is confined to the interval (−π/2, π/2), it is clear
that they do not form a group.

A new parameter is then chosen (the angle of rotation α), thereby extending the transformations to a full group, and the
properties of the latter are investigated. Two of them may be employed to rule out this case:
• (pre-) causality is broken;
• the composition of two positive velocities may yield a negative velocity.

(See also Ref.11 , Counterexample 4.) Some Authors (see references in10 ) use the second of these: they consider such a paradox
to be physically unacceptable. However, in Ref.10 this argument is rejected, since in principle the phenomenon may not be
excluded—mutatis mutandis, it is even predicted in the theory of tachyons—and we do agree with this opinion. What is left out
is (pre-) causality, which is indeed a reasonable assumption (from the physical point of view). On the other hand, imposing v as
a parameter is unavoidable, and eliminates the need for such an assumption. √ √
Let us prove our statements. Write k = −K (K > 0). We may define the novel variables ξ = t/ K , ξ 0 = t 0 / K ; from
formulae (4), we find:
 0  
ξ ξ
= Ω (α) , (16)
x0 x

where
 
cos α sin α
Ω (α) = (17)
− sin α cos α

and α is a solution to the couple of equations


(
cos α = γ (v) (18a)

sin α = K vγ (v) (18b)

(equation (11) yields 1 + K v2 γ 2 (v) = 1, so there are indeed (infinite) solutions to equations (18)). Note that, since γ (v) is


positive, there is a (unique) solution in the (open) interval ]−π/2, π/2[. We may also set a (real) value for α and see whether the
matrix Ω (α) given by formula (17) might represent a physical transformation (i.e., there be a (real) v such that (18) be satisfied);
this is only the case if cos α is positive, which amounts to say that α belongs to ]−π/2, π/2[ (except for a multiple of 2π). It is
easily seen that (17) represents a (counterclockwise) rotation of the axes by an angle equal to α: this geometrical interpretation
immediately yields Ω (α2 ) Ω (α1 ) = Ω (α1 + α2 ) for any (real) α1 , α2 (the composition of two successive rotations by α1 and
α2 respectively amounts to a single rotation by α1 + α2 ). By (mathematical) induction, we have Ωn (α) = Ω (nα) (where n is
any positive integer).
Now, we state that the only possible boost is the identical one (v = 0), i.e. Γ = {0}. Indeed, if there were a boost with
a velocity v 6= 0 then it would be possible to devise, by composition, a transformation which is not a boost, and this would
contradict condition i). This is most easily seen by expressing our hypothetical boost as a rotation by an angle α (∈ ]−π/2, π/2[):
since v 6= 0, equations (18) yield α 6= 0, so we may find a (positive integer) n s.t. |nα| > π/2; if n is the smallest solution to this
inequality, we also have |nα| < π (since |α| < π/2). As a consequence, Ω (nα) does not represent a physical transformation,
so the composition of n boosts with velocity v (which is represented by Ωn (α) = Ω (nα)) is not a boost. From the point of
view of group theory, this means that the cyclic group generated by any non-identical transformation is not included into the set
of physical transformations, which therefore cannot be a group (unless it reduce to the trivial group whose only element is the
identity).
As a final remark, note that the Lorentz transformations include as a special case the Galilean transformations (which are
obtained for c → ∞) and also the case k < 0 (which corresponds to c → 0).

6/7
Conclusion
In the original derivation by Einstein, the transformations of coordinates are formally identical to our formulae (14); however,
the parameter c is identified with the speed of light. Our discussion has led us to the explicit form of the transformations
without imposing the Principle of the Constancy of the Velocity of Light, but c remains unknown (it should be determined by
experiment). Whatever be the actual value of c, the resulting transformations do form a group and imply (pre-) causality: this
shows that (pre-) causality is already implemented by our three assumptions—homogeneity of space and time, isotropy of
space and the Principle of Relativity—and there is no need to introduce it as a further assumption.

Acknowledgments
I wish to thank Prof. Tullio Weber and Prof. Giovanni De Ninno for valuable discussion and advice.

Competing financial interests


The author declares no competing financial interests.

References
1. Einstein, A. Zur elektrodynamik bewegter körper. Annalen der Physik (Leipzig) 322, 891 (1905).
2. Lorentz, H. A., Einstein, A., Minkowski, H. & Weyl, H. The Principle of Relativity: A collection of original memoirs
on the special and general theory of relativity (Methuen and Company, London, 1923). With notes by A. Sommerfeld.
Translated by W. Perrett and G. B. Jeffery.
3. Einstein, A. The Collected Papers of Albert Einstein, Vol. 2, The Swiss Years: Writings, 1900–1909. English translation
(Princeton University Press, Princeton, 1989). By A. Beck, translator; P. Havas, consultant.
4. von Ignatowsky, W. Einige allgemeine bemerkungen zum relativitätsprinzip. Verhandlungen der Deutschen Physikalischen
Gesellschaft 12, 788 (1910).
5. von Ignatowsky, W. Einige allgemeine bemerkungen über das relativitätsprinzip. Physikalische Zeitschrift 11, 972 (1910).
6. von Ignatowsky, W. Das relativitätsprinzip. Arch. der Math. und Physik 17, 1 (1911).
7. von Ignatowsky, W. Das relativitätsprinzip. Arch. der Math. und Physik 18, 17 (1911).
8. von Ignatowsky, W. Eine bemerkung zu meiner arbeit: ”einige allgemeine bemerkungen zum relativitätsprinzip”.
Physikalische Zeitschrift 12, 779 (1911).
9. Lalan, M. V. Sur les postulats qui sont à la base des cinématiques. Bull. de la Soc. Math. de France 65, 83 (1937).
10. Berzi, V. & Gorini, V. Reciprocity principle and the lorentz transformations. J. Math. Phys. 10, 1518 (1969).
11. Lévy-Leblond, J.-M. One more derivation of the lorentz transformation. Am. J. Phys. 44, 271 (1976).
12. Liberati, S., Sonego, S. & Visser, M. Faster-than-c signals, special relativity, and causality. Annals Phys. (N.Y.) 298, 167
(2002).
13. Coleman, B. A dual first-postulate basis for special relativity. Eur. J. Phys. 24, 301 (2003).
14. Sonego, S. & Pin, M. Foundations of anisotropic relativistic mechanics. J. Math. Phys. 50, 042902 (2009).
15. Pelissetto, A. & Testa, M. Getting the lorentz transformations without requiring an invariant speed. Am. J. Phys. 83, 338
(2015).
16. Brown, H. R. Physical Relativity: Space-time structure from a dynamical perspective (Oxford University Press, Oxford,
2005).
17. Eisenberg, L. J. Necessity of the linearity of relativistic transformations between inertial systems. Am. J. Phys. 35, 649
(1967).

7/7

You might also like