Thermodynamics in

Variable Speed of Light Theories

Juan Racker1,3∗, Pablo Sisterna2†, and Hector Vucetich3‡
1
CONICET, Centro Atómico Bariloche, Avenida Bustillo 9500 (8400) Argentina
2
Facultad de Ciencias Exactas y Naturales, Universidad Nacional de Mar del Plata,
Funes 3350 (7600) Mar del Plata, Argentina
3
Facultad de Ciencias Astronómicas y Geofı́sicas, Universidad Nacional de La PLata,
Paseo del Bosque S/N (1900) La Plata, Argentina

November 27, 2008

Abstract
The perfect fluid in the context of a covariant variable speed of light (VSL)
theory proposed by J. Magueijo is studied. On the one hand the modified first
law of thermodynamics together with a recipe to obtain equations of state are
obtained. On the other hand the Newtonian limit is performed to obtain the
non-relativistic hydrostatic equilibrium equation for the theory. The results ob-
tained are used to determine the time variation of the radius of Mercury induced
by the variability of the speed of light (c), and the scalar contribution to the lu-
minosity of white dwarfs. Using a bound for the actual change of that radius and
combining it with an upper limit for the variation of the fine structure constant,
a bound on the time variation of c is set. An independent bound is obtained
from luminosity estimates for Stein 2015B.

1 Introduction
There are several very different motivations for studying the variation of fundamental
constants. The coincidence of large dimensionless numbers arising from the combina-
tion of different physical constants led Dirac to propose the Large Number Hypothesis
and predict a time variation of them [1], [2]. Theories with varying constants can also
be a way of implementing Mach’s principle. Extra-dimensional theories like superstring
∗
racker@cab.cnea.gov.ar
†
sisterna@mdp.edu.ar
‡
vucetich@fcaglp.unlp.edu.ar

1
or Kaluza Klein theories reduce in the low energy limit to effective theories in which
the fundamental constants may vary in space and time.
Although no variation has been found in most experiments and observations per-
fomed up to date, the results from analysis of spectra from high redshift quasar absorp-
tion systems remain controversial. Some works have reported a variation in the fine
structure constant [3], [4] , [5], [6], [7], [8], while other studies give null results [9], [10],
[11], [12]. Besides the motivations mentioned above, variable speed of light theories
(VSL) are interesting because they could solve several cosmological puzzles [13], [14],
[15].
In this work we study the thermodynamics and Newtonian limit of the varying speed
of light theory developed by J. Magueijo [16]. After a summary of this VSL theory
(section 2) and of a lagrangian approach to describe perfect fluids (section 3), the first
law of thermodynamics and a recipe for obtaining equations of state are derived (section
4). It is shown that the field associated with the variation of the speed of light (c)
can formally be considered as a new thermodynamical variable. Regarding this point
we note that this field has two properties that are not common in thermodynamical
variables: local universality and long scale variation. This has been explained in a
previous work in the context of another theory with variation of physical constants
[17]. In section 5 we perform the newtonian limit and in section 6 we apply all results
to study the evolution of the radius of Mercury and derive a bound on the time variation
of c. Section 7 devotes to the estimation of the scalar contribution to the luminosity
of a white dwarf, obtaining a stringent upper bound for the variation of c. In section
8 we state our conclusions and we leave for an appendix some results concerning the
space and time dependence of the scalar field.

2 Brief description of the VSL theory

In the covariant and locally Lorentz invariant VSL proposed by Magueijo c plays three
different roles:
a) c is a dynamical field:
The spacetime variations of c are represented by an adimensional scalar field ψ,
so that
c = c0 eψ , (1)
where c0 is a constant.
The General Relativity (GR) action is modified and becomes:
√
Z  
4 aψ µ 16πG bψ
I = d x −g e (R − 2Λ − κ∇µ ψ∇ ψ) + e Lm . (2)
c04
The metric is taken to be ηµν = Diag(−1, 1, 1, 1). Lm is the matter lagrangian
and κ, a and b are three parameters of the theory. We will take a − b = 4 as
in [16]. The matter lagrangian is required to have no explicit dependence on c
(minimal coupling condition). This leads to the second role played by c.

2
 b) c parametrizes the variations of the other “constants”: The minimal coupling
condition fixes the scaling with c of all the lagrangian parameteres up to the
~(c) dependence, which is taken to be ~ ∝ cq−1 . q is the forth parameter of the
theory. For example, since the Compton wavelength appears in the lagrangian
of a quantum particle, the dependence on c of the mass will be m ∝ ~/c ∝ cq−2 .
In a similar fashion it can be determined that the charge of a quantum particle
(e), the Bohr’s radius (rb ) and the fine structure constant (α) are proportional
to cq , c−q and cq respectively.

c) c is an integrating and conversion factor: The theory is covariant and locally

Lorentz invariant in a generalized way explained in [16]. The key point is the use
of an x0 coordinate, instead of time, in all geometrical formulas. The main differ-
ence with constant c theories is that local measurements of space and time (dx and
dt) aren’t generally differentials of a coordinate system (the partial derivatives
don’t commute). Nevertheless it is always possible to find integrating factors such
that dt ψ β and dx ψ β−1 are perfect differentials. β is another parameter which,
however, won’t appear in the equations of our work, because the calculations will
be done using the x0 coordinate.

Varying the action (2) with respect to the metric and ψ leads to the equations:
 
8πG 1 δ
Gµν + Λgµν = 4 Tµν + κ ∇µ ψ∇ν ψ − gµν ∇δ ψ∇ ψ +
c 2 (3)
−aψ aψ aψ


e ∇µ ∇ν e − gµν e ,

8πG
1 dΛ̄
ψ + a∇µ ψ∇µ ψ = aT − 2aρ Λ c 2
− 2bL m + . (4)
c4 (2κ + 3a2 ) κ dψ
T µν is the matter stress energy tensor and T is its trace. ρΛ is the mass density
Λc2
corresponding to the cosmological constant, ρΛ = 8πG and Λ̄ is a linear combination of
the matter and geometric cosmological constants, Λ̄ = Λ + 8πGc4
Λm .
After applying the Bianchi identities to (3) and using (4), an equation for the
divergence of T µν is obtained:

T ν µ;ν = −ψ;ν T ν µ b + ψ;µ bLm − Λm;µ . (5)

Note that matter energy is conserved only when b = 0, a = 4. In all other

cases there is exchange of energy between matter and the ψ field.
The presence of the ψ field can modify the law of conservation of the number of
particles and the normalization condition for the four-velocity, e.g. it is found that for a
classical particle U 2 = U02 (c/c0 )−b 6= −1 [16]. Besides, the energy density and the total
energy of a body in hydrostatic equilibrium will also vary if c isn’t constant (we must
distinguish between energy density and total energy because the size of a body can
change in time as a result of the variation of c). In addition, the c dependence of the
mass is different for a classical and a quantum particle. Finally, the matter lagrangian,

3
which is not unique, appears in the equations of the theory. To take into account all
these effects and ambiguities it is convenient to introduce four new parameters, q1 , q2 , q3
and q4 :

Generalized normalization
cq1 U µ Uµ = cq01 U02 = C (6)
of the four-velocity.
Generalized conservation
(ncq2 U ν );ν = 0 (7)
of particle number.
c dependence of the energy density of
ρ = ρ0 eq3 ψ (8)
a body in hydrostatic equilibrium.
c dependence of the total energy of
U = U0 eq4 ψ (9)
a body in hydrostatic equilibrium.

The “ 0 ” subscript denotes the value of these quantities when ψ = 0.

3 The perfect fluid lagrangian

The task of obtaining a lagrangian for the perfect fluid is not a trivial one due to the
constraints imposed by the normalization of the velocity and the conservation of the
number of particles. A. H. Taub gave one in 1954 [18] and another one was given by B.
F. Schutz in 1970 [19], using a different but equivalent approach. Note that equivalent
lagrangians in usual theories (i.e. differing in a divergence) may not be equivalent in
Magueijo’s one, because of their explicit appearance in the evolution equations. We
will use Schutz’s lagrangian. A brief summary of his approach is given in the following
section.

3.1 Schutz’s lagrangian

Schutz uses a formulation of relativistic hydrodynamics based on the utilization of 6
potentials to represent the velocity:

Uν = µ−1 (φ,ν + αβ,ν + θs,ν ) . (10)

where µ and s are the specific enthalpy (enthalpy per unit mass) and the specific
entropy respectively. The physical meaning of the other potentials is also explored by
Schutz.
The perfect fluid action is:
Z  
16πGp
I= R+ (−g)1/2 d4 x ,
c04

where p is the pressure of the fluid. Then the perfect fluid lagrangian is: Lm = p. The
following steps must be followed to vary the action:

4
 1. Choose an equation of state for the fluid and write it in terms of µ and s: p =
p(µ, s).

2. When varying the action make use of the thermodynamic relation dp = ρm dµ −

ρm T ds (ρm is the rest mass density).

3. Define the four-velocity vector field Uν in terms of 6 scalar potentials: Uν =

µ−1 (φ,ν + αβ,ν + θs,ν ).

4. The normalization of U is taken into account before starting the variations. This
is done expressing µ in terms of φ, α, β, θ, s, g µν : µ2 = −g µν (φ,µ + αβ,µ + θs,µ )
(φ,ν + αβ,ν + θs,ν ). So the independent variables are φ, α, β, θ, s and g µν . Any
quantity appearing in the action must be considered a function of these variables.
The Euler-Lagrange equations become:
8πG
Gµν − [(ρ + p)Uµ Uν + pgµν ] = 0 ,
c04
(ρm U ν );ν = 0 ,
U ν s,ν = 0, U ν θ,ν = T, U ν β,ν = 0 and U ν α,ν = 0 .

The stress energy tensor is:

√
µν 2 δ ( −gLm )
T =√ = (ρ + p)U µ U ν + pg µν . (11)
−g δgµν

It’s important to note that (ρ + p), Uµ , p, ρm and T were defined in terms of φ, α, β, θ, s

and g µν after performing the variations:

p = p(µ, s) (equation of state),

   
∂p −1 ∂p
ρm = , T = ,
∂µ s ρm ∂s µ
(ρ + p) = ρm µ, Uν = µ−1 (φ,ν + αβ,ν + θs,ν ) .

3.2 Use of Schutz’s lagrangian in Magueijo’s theory

Using Schutz’s lagrangian in the action (2) and varying it with respect to φ, α, β, θ and
s we obtain :

(ebψ ρm U ν );ν = 0, (12)

U ν s,ν = 0, (13)
U ν θ,ν = T, (14)
U ν β,ν = 0, (15)
U ν α,ν = 0. (16)

5
 Varying Lm with respect to g µν leads to:

T µν = (ρ + p)U µ U ν + pg µν . (17)

ρ, p, ρm , U µ are defined from φ, α, β, θ and s exactly in the same way as in Schutz’s

theory. In the VSL theory they can depend on ψ, but will coincide with the usual
quantities in the case ψ = 0. Besides, with these definitions of ρ, p and U µ , the stress
energy tensor in the VSL theory has the same form as the usual one for a perfect fluid
with energy density ρ and pressure p (both quantities being measured in a momentarily
comoving reference frame). These facts make it reasonable to consider ρ, p, ρm and U µ
as the energy density, pressure, rest mass density and four velocity of the perfect fluid
in the VSL theory that is being studied.
Note that by definition U µ Uµ = −1, so q1 = 0 . This is different from Magueijo’s
result for a classical particle, ebψ U µ Uµ = constant, where the definition of the velocity
µ µ
is U µ = dxdλ
= dxcdτ
. Although there’s no contradiction with this, one has to be careful
with the interpretations given to U µ .

4 Thermodynamics
4.1 First law
When energy is conserved, the divergence of T µν (Eq.(5)) is zero and the first law of
thermodynamics is obtained projecting along U µ . We will do the same here but using
cq1 U µ as the projector. Although using Schutz’s lagrangian leads to work with a four-
velocity normalized to −1 (q1 = 0), the calculations in this section will be done with
an arbitrary q1 . The motivation is to obtain equations valid even when that condition
is not satisfied.
The results of this section will be applied to systems whose scales are much smaller
than cosmological scales, so we take Λm = 0 1 . Projecting T µ;ν ν
along cq1 U µ and using
(5), (17), (6) and (7) together with Schutz’s lagrangian leads to 2

cq1
 
(ρ + p) q
1

dρ + 1 + dp − dn = (ρ + p) + q2 − b dψ . (18)
C n 2

It is convenient to express this relation in terms of the specific thermodynamic variables,

V 1 U S
v= = , u = ρv = , s = (V , U and S are the total volume, total energy and
N n N N
total entropy of a system containing N particles):

cq1
q   
1
du + pdv − v(ρ + p) + q2 − b dψ + v 1 + dp = 0 . (19)
2 C
1
On the other hand, the cosmological constant is important for the evolution of ψ.
q1
dΛm
If Λm 6= 0 but depends only on ψ, the equation will be valid after adding the term − cC
2
dψ dψ
to the right hand side.

6
 The first two terms are the only ones appearing in GR. The third term shows
that ψ formally plays the role of a new thermodynamic variable (changes in ψ cause
changes in the internal energy). The fourth term involves the pressure, which is not an
independent variable (up to this point the independent variables are taken to be v and
ψ). This expression is valid for thermodynamical processes which don’t involve heat
transfer. In the general case it becomes

cq1
q   
1
du + pdv − v(ρ + p) + q2 − b dψ + v 1 + dp = dQ . (20)
2 C

This is the modified first law of thermodynamics in the VSL theory.

Incorporating Caratheodory’s principle it can be shown that there exists an inte-
grating factor for dQ in (20) [20]. Defining T1 as the integrating factor and ds = dQ
T
,
(20) becomes:

cq1
  q 
1
du + v 1 + dp + pdv − v(ρ + p) + q2 − b dψ = T ds . (21)
C 2

s and T are two thermodynamical variables. s will be identified with the specific
entropy of the system. q
q c 1
It is convenient to introduce the function f (ψ) = 1 + cC1 = 1 + C0 eq1 ψ . We take
as the time of reference the present epoch, so U02 = −1 (usual normalization of the
four-velocity) and cq01 = cqtoday
1
. Then C = −cq01 and f (ψ) = 1 − eq1 ψ .
After choosing s, v and ψ as the independent variables and writing p in terms of
them, an expression for the first law involving only state variables is obtained:

   
∂p ∂p
du = − p + vf (ψ) dv + T − vf (ψ) ds +
∂v ∂s
(22)
 q 
1
 ∂p
(u + pv) + q2 − b − vf (ψ) dψ
2 ∂ψ

4.2 Equations of state

The first partial derivatives of the specific internal energy can be obtained directly from
(22). There are three different equalities between the mixed partial derivatives which
impose some restrictions on the functional dependence of p and T on ψ:
∂p ∂T
(f (ψ) − 1) = , (23)
∂s ∂v
∂p ∂p
= b1 v , (24)
∂ψ ∂v

7
 ∂T ∂T
= b2 T + b1 v . (25)
∂ψ ∂v
with b1 = q21 − q2 + b and b2 = q21 + q2 − b.
Several observations corresponding to different epochs in the history of the universe
show that the α variation has been very small (or zero) [21], so the field ψ must also
be very small. Then, it makes sense to express the pressure as a power expansion in ψ
3
: ∞
X
p= pk (v, s)ψ k . (26)
k=0

After replacing this series in (24) and equaling terms with the same power of ψ a
recurrent formula for the coefficients pk is obtained:
b1 v ∂pk (v, s)
pk+1 (v, s) = . (27)
k + 1 ∂v
where p0 (v, s) is the pressure as a function of v and s for ψ = 0 and so it is obtained
from the usual theories in which c is constant.
Working to first order in ψ we arrive at the following expression for the functional
dependence of p on v, s and ψ:
∂p0
p ' p0 (v, s) + b1 v ψ. (28)
∂v

5 Newtonian limit
The Newtonian limit of this VSL theory can be obtained following the same steps as
those used in GR. We will work to first order in ψ. Although in the Newtonian limit
time derivatives are negligible with respect to the spatial ones, special considerations
are needed in the case of ψ. For example, if the spatial extension of a system is small
compared to the scales associated with ψ variations (cosmological scales) and one is
interested in following its time evolution, time derivatives will be more interesting than
the spatial ones. This will be the case in the following sections, where we will determine
the evolution of the planetary radii and the luminosity of white dwarfs. As explained
before, we haven’t taken into account the cosmological constant.
The condition of weak gravitational field allows to choose nearly Lorentz coordinates
in which
gµν = ηµν + hµν , with |hµν |  1 . (29)
Using (3), Lm = p and the fact that a∇0 ∇0 ψ is negligible (see the appendix), the
following gravitational potential equation is obtained:

8π G̃
−∇2 h00 = T00 . (30)
c4
3
The ψ dependece of the temperature can be treated in a similar fashion.

8
  
a2
Here G̃ = G 1 + 2κ+3a2
is an effective gravitational constant. In the non relativistic
limit of GR h00 is identified with −2 cφ2 to arrive at Newton’s gravitational potential
equation. Here we will go on working with h00 due to the presence of the c2 factor in
that identification.
The Euler’s equations for this theory are obtained projecting (5) perpendicularly
to U µ , using the projector g µα + U µ U α (here we have taken q1 = 0 from the beginning):
1
− T00 h00,i + p,i + ρU ν Ui,ν + ρU0 U j g0j,i = 0 . (31)
2
For quasi statical situations (Ui ≈ 0) the equation becomes:
1
T00 ∇h00 = ∇p . (32)
2
The hydrostatic equilibrium equation follows after combining (30) and (32). For a
system with spherical symmetry the result is:

dp G̃ ρ(r)U (r)
=− 4 . (33)
dr c r2
d
where U (r) is the total energy inside the sphere of radius r. dr must be understood as
a spatial derivative at constant time. To get this equation it is necessary to consider c
R r 4πr 2
as a constant in the integral 0 c4 ρ dr. This can be done because the spatial variations
of ψ are negligible in non-cosmological scales, as is demonstrated in the appendix.

6 Evolution of the radius of Mercury and a bound

for the variation of c
The radius of a planet is determined with the hydrostatic equilibrium equation to-
gether with an equation of state and boundary conditions. The presence of ψ in these
equations causes in general time variations of the radius. The actual change in size of
several bodies of the solar system have been estimated using different topographical
observations. For Mercury there’s a stringent bound: its radius hasn’t changed more
than 1 kilometer in the last 3.9 × 109 years [22]. This fact will allow to obtain a bound
for the temporal variation of ψ.
In 8.1 we will show that the hydrostatic equilibrium equation is equivalent to an-
other equation in which the temporal dependence resides only in the gravitational
constant. Then it will be possible to use the results of the work of McElhinny et al.
and give a bound for the quantity f (qi )ψ̇, where f (qi ) is a function of the parameters
b1 , q3 and q4 (the dot denotes derivative with respect to time). In 8.2 those parameters
will be expressed in terms of the parameters originally defined in [16]. Finally, in 8.3
the bound obtained for f (qi )ψ̇ will be combined with a bound for the variation of the
fine structure constant to obtain an upper limit for |ψ̇|.

9
6.1 Transformation into a variable G theory
To solve the hydrostatic equilibrium equation it is necessary to have an equation of
state relating the pressure and density. It can be obtained replacing the corresponding
equation of state for the constant c case in equation (28). For Mercury it’s sufficient
to work with a linear equation in ρ [22]:
 
ρ0
p0 = Ksur −1 . (34)
ρ0 sur

The quantities with a “0” subscript correspond to ψ = 0 and “sur” indicates evaluation
in the surface of Mercury. Ksur is the superficial compressibility 4 .
m c2
Using (28), (34) and the non relativistic expression ρ0 ≈ 0v 0 (m0 is the average
mass of a particle), the equation of state to be used is obtained:
 
ρ0 ρ0
p = Ksur − 1 − b1 Ksur ψ . (35)
ρ0 sur ρ0 sur
∂ψ
Replacing this equation in (33), using the expression for ∂r
given in the appendix
and the definition of q3 and q4 we get:

dp0 ρm0 (r)M0 (r) Ksur a ρm0 (r)M0 (r)

= −G̃ e(b1 +q3 +q4 −4)ψ 2
− 2b1 2
eb1 ψ G . (36)
dr r ρ0 sur 2κ + 3a r2

ρm0 (r) is the mass density for ψ = 0 and M0 (r) is the mass contained within radius r
(also for ψ = 0). The second term of the right hand side is small compared to the first
one5 , so it can be multiplied by e(q2 +q4 −4)ψ (which is equal to 1 + O(ψ)). The equation
becomes:
dp0 ρm0 (r)M0 (r)
= −Ḡ e(b1 +q3 +q4 −4)ψ . (37)
dr r2
Ḡ is the final effective gravitational constant:

a2
 
Ksur a
Ḡ = G 1 + + 2b1 . (38)
2κ + 3a2 ρ0 sur 2κ + 3a2

The last term within the parenthesis depends on the particular properties of the body
and could have, in principle, observational consequences. Nevertheless, it’s value is
too small relative to the second (and body independent) term for it’s effect to be
measurable.
The method used to set a limit on the actual time variation of the radius of Mercury
is based on the observation of surface features [22]. A homologous change (R ∝ M 1/3 )
4
In [22] a quotient between mass densities is used instead of ρ0ρsur
0
. In the Newtonian limit they
2
are equal since ρenergy ≈ ρmass c .
5
Assuming b1 , a = O(1), the quotient between these terms is no larger than approximately ρK0sur
sur
.
dynes dynes Ksur
Ksur ≈ 1012 cm2 and ρ0 sur = ρm0 sur c02 ≈ 4 × 1021 cm2 [23], then ρ0 sur ≈ 10−9 .

10
can’t be detected through this procedure since it scales all linear dimensions equally.
Neither can be observed a variation R ∝ rb ( rb is Bohr’s radius ), since a change in
rb implies a change of all macroscopic dimensions in the same proportion. For these
reasons it’s necessary to make a change of variables:
r M (r) M0 (r)
r∗ = , M∗ = = , (39)
rb M 1/3 M M0
where M is the total mass of Mercury.
Given that m ∝ cq−2 and rb ∝ c−q , Eq. (37) becomes
2/3
dp0 M ρm0 (r∗ )M0∗ (r∗ )
∗
= −Ḡ(t) 0 , (40)
dr rb 0 r∗ 2
with

Ḡ(t) = Ḡ e(b1 +q3 +q4 −4+2/3q+2/3)ψ(t) . (41)

rb 0 is Bohr’s radius for ψ = 0.
This equation shows that the VSL theory that is being studied is equivalent, with
respect to the hydrostatic equilibrium of Mercury, to a theory in which the only “con-
stant” that varies is G. The variation in the radius of Mercury (R) produced by the
variation in G can be parametrized as [22] 6 :
1 dR δ dḠ(t)
=− , (42)
R dx0 Ḡ(t) dx0
or equivalently
1 dR δ dḠ(t)
=− . (43)
R dt Ḡ(t) dt
δ is generally a function of Ḡ and M . Using models of Mercury, McElhinny et al.
obtained for that planet δ = 0.02 ± 0.005.

6.2 Specification of the parameters

We will express the parameters b1 , q1 , q2 , q3 and q4 that have been introduced in this
work in terms of the parameters q and b of the VSL theory. In our approximation
M = N MN where N is the total number of nucleons. Equation (7) requires that
N ∝ exp−q2 ψ while (12) implies that M ∝ exp−bψ and from the discussion in section
2-b, we see that M ∝ exp [(q − 2 − q2 )ψ] and consequently q2 = b + q − 2. From (8)
q3 = q−2+3q = 4q−2 while from (9) q4 = −q2 +q−2. Also we have b1 ≡ q1 /2−q2 +b =
−q + 2 and as we explained before we consider the q1 = 0 case.
Finally from (41) we conclude that
2 2 11 10
Ḡ(t) = Ḡ exp(b1 +q3 +q4 −4+ 3 q+ 3 )ψ = Ḡ exp( 3 q−b− 3 )ψ . (44)
6
We call “δ” the parameter called “α” in [22] to avoid confusion with the fine structure constant.

11
6.3 Bound for ċ/c
Replacing the previous expression for Ḡ(t) in (43) one gets:
 
11 10 1 Ṙ
q−b− ψ̇(t) = − ' 0 ± 5 × 10−12 y−1 , (45)
3 3 δ R

where y−1 = 1/year. We have taken δ = 0.02 ± 0.005 and ∆R R

= 0 ± 0.0004 [22]. ∆R
9
corresponds to a time interval approximately equal to 3.5 × 10 years.
This result can be combined with bounds for α̇/α that have been obtained using
atomic clocks. We can use e.g. the one obtained in [24]):
α̇
= (4.2 ± 6.9) × 10−15 y−1 . (46)
α
In this paper we will consider the b = 0 case, which gives an upper bound for all
non negative values of b. In the VSL theory α̇α = q ψ̇. On the other hand, (45) can
be written − 103
ψ̇ = 0 ± 5 × 10−12 y−1 − 11
3
q ψ̇. Comparing with (46) we see that
10 −12 −1
− 3 ψ̇ ≈ 0 ± 5 × 10 y . The conclusive result is:
ċ
= ψ̇ = 0 ± 2 × 10−12 y−1 . (47)
c
Multiplication by the Hubble time ( H0−1 ) results in a bound for the adimensional
quantity ψ 0 = H0−1 ψ̇ : 7
ψ 0 = 0 ± 3 × 10−2 . (48)

7 White dwarfs luminosities and the scalar field

White dwarfs are excellent objects to test any energy injection from a scalar field
[25, 26]. This is due both to their low luminosity, as well as their extremely high
heat conductivity, making all energy microscopically released to enhance the total
luminosity. Most of them are adequately described by Newtonian physics and a zero
temperature approximation, the latter hypothesis providing a politrope type equation
of state. Using the subscript 0 to denote quantities without the presence of the scalar
field, we write the politrope equation as p0 = K0 ργ . Given (28), we have
p0
p = p0 (ρ) − b1 ρ ψ. (49)
ρ
so we can write
p = K0 (1 − b1 γψ)ργ , (50)
and we recover a politrope equation of state with a new constant K0 → K = K0 (1 −
b1 ψγ). The Lane-Emden function with politrope index (γ − 1)−1 still applies, and
7
we have taken H0−1 ≈ 1.5 × 1010 years.

12
consequently the expressions for the radius, mass and internal energy of the star will
be the same in terms of the effective constant K:
3γ − 4 GM 2
E=− (51)
5γ − 6 R
 1/2
Kγ
R= ρ(γ−2)/2
c ζ1 (52)
4πG(γ − 1)
 3/2
Kγ
M = 4πR (3γ−4)/(γ−2)
ζ12 |θ0 (ζ1 )| (53)
4πG(γ − 1)
where ρc is the central density and ζ1 is the root of the Lane-Emden function.
From (44) we can write the dependence of the internal energy on ψ:

E ∝ expψf (q,b,γ) (54)

where
11 10 γ−2 5 3 11
f (q, b, γ) ≡ q − 3b − + ( b − bγ − q + 5). (55)
3 3 3γ − 4 2 2 2
The equation for the luminosity becomes

L = −Ė = −f (γ, q, b)E ψ̇. (56)

The stars considered are well described by the non relativistic value γ = 5/3, in which
case f (q, b, γ) = 11
2
q − 3b − 5.
We use again the upper bound (46) for the present time variation of α obtaining
the following expression conveniently written as
L 11 α̇
ψ̇ = + (57)
(3b + 5)E 2(3b + 5) α
5 11
= ψ̇0 ± 1.1 × 10−14 y −1 ,
3b + 5 2(3b + 5)

where
(L/L ) 1
ψ̇0 ≡ , (58)
(E/E0 ) 5τ
and
3 GM 2
E0 = − , (59)
7 R
is the would be internal energy of the sun were it described by a newtonian γ = 5/3
politrope. L is the solar luminosity and
E0
τ = ' 1.32 × 107 y, (60)
L

13
 Object M/M R/R L/L E/E0 ψ̇0 |ψ̇| ≤
Procion B 0.602 0.0123 5.8 × 10−4 29.5 3.0 × 10−13 3.1 × 10−13
40 Eri B 0.501 0.0136 0.014 18.5 1.2 × 10−12 1.0 × 10−10
Stein 2015B 0.66 0.011 3.1 × 10−4 39.6 1.2 × 10−13 1.3 × 10−13

Table 1: Data and limits for selected white dwarfs

is the Kelvin-Helmholtz solar contraction time scale. The following table shows upper
bounds for ψ̇ in y −1 again for b = 0:
Stein 2015B provides the stronger bound. Using a value for the Hubble time H0−1 ∝
1.5 × 1010 y we obtain
1 ċ 1
= ψ̇ = 0 ± 2.0 × 10−3 . (61)
H0 c H0
We see that white dwarf physics provide the strongest constraints on VSL theories near
the present epoch.
Although equations (45) and (57) have only two unknown variables ψ̇ and b, they
are almost useless without some further assumption. Furthermore, after some algebraic
manipulation the following equation for b results and using the largest value of ψ̇ in
the table:
3b + 10
−δ < < δ ∼ 1. (62)
9b + 15
This condition is satisfied for almost all values of b, namely b ∈
/ (−2.08, −0.83) (the
forbidden interval). Therefore the value b = −10/3 which make Eq. (45) uninformative
about ψ̇ is allowed. Thus we need to make some assumption regarding b, as we did
above.

8 Conclusions
We have obtained the equations for a perfect fluid in the non-relativistic limit and
the first law of thermodynamics in the context of the covariant VSL theory proposed
by J. Magueijo. The field ψ can formally be considered as a new thermodynamical
variable. It has been shown how to obtain the equation of state for the pressure when
the equation for constant c is given.
The non-relativistic hydrostatic equilibrium equation has the usual form with the
gravitational constant G replaced by an effective constant. The different variables
(pressure, mass, mass density) depend on the ψ field and so the radius of a planet
should vary in time. Using bounds for the variation of the radius of Mercury and the
fine structure constant we have set limits on ċ/c : ċ/c = ψ̇ = 0 ± 2 × 10−12 y−1 . The
most interesting thing about this result is that it gives a bound for ψ̇, whereas the
known limits for the variation of α and e lead to bounds for the product q ψ̇.
Under the same Newtonian approximation we obtained the dependence of the lu-
minosity of a white dwarf on the time variation of the scalar field. The bound obtained
is more stringent than the planetary radius bound by an order of magnitude.

14
 The b = 0 assumption suggests a null coupling of the scalar field with matter.
However, the q 6= 0 assumption implies a quantum coupling between ψ and matter,
not explicitly shown in the action (2). Of course this and other issues such as the
microscopic origin of the energy exchange between the scalar field and ordinary matter
as well as whether all the energy injected by ψ on a star is radiated away or not, deserve
further work. This we leave for future communications.

9 Appendix: spatial and temporal behavior of ψ

After some simplifications valid in the Newtonian limit, the equation for ψ (in the case
a 6= 0) becomes:
8πG
ψ = 4 aT . (63)
c (2κ + 3a2 )
Since ψ is small, c4 can be replaced by c04 in the right hand side of the equation (this
is the first step of an iterative process of resolution):
8πG
ψ = aT . (64)
c04 (2κ+ 3a2 )
T can be written as the sum of a spatial part Ts (corresponding to Mercury) and a
temporal one Tt (cosmological term). Then ψ can also be separated into spatial (ψs )
and temporal (ψt ) components. The metrics to be used are the quasi Minkowskian and
the cosmological (Robertson-Walker) for the spatial and temporal parts respectively.

9.1 Spatial behavior

Using that T ' ρ = energy density of Mercury, the spatial part of equation (64) can
be written:  
2 1 d 2 dψs 8πG a
∇ ψs = 2 r = 4 ρ. (65)
r dr dr c0 2κ + 3a2
After integrating one arrives at:
dψs 2G a U (r)
= 4 . (66)
dr c0 2κ + 3a r2
2

This is one of the formulas that has been used in our work. It can help us to obtain
an estimation of the variation of ψ inside Mercury:
 
2a 4 1 G 2
∆ψ ≈ π ρ¯m R , (67)
2κ + 3a2 3 2 c02
where ρ¯m and R are the average density and the radius of Mercury respectively. The
first factors are presumably O(1) so:
G
∆ψ ≈ ρ¯ R2 ≈ 10−11 .
2 m
(68)
c0

15
This number is small compared to the bound that is obtained for the variation of ψ
in the last 3.9 × 109 years (the relevant period of time for our study of Mercury) and
besides the temporal and spatial behavior of this field can be separated. This justifies
the steps of our analysis in which ψ was considered constant inside Mercury.

9.2 Temporal behavior

The temporal part of equation (64) is:

8πG a
ψt = 2
ρmU , (69)
c0 2κ + 3a2

where ρmU is the average density of the universe. Using the Robertson-Walker metric
it becomes:
8πG a ȧ(t)
∇0 ∇0 ψ = 2 2
ρmU + 3 ψ̇, (70)
c0 2κ + 3a a(t)
where a(t) is the scale factor of the metric and a is one of the parameters of the
VSL theory. In the way towards obtaining the gravitational potential equation in the
Newtonian limit, the equation G00 = R00 − 21 R = 8πG
c4
T00 + a (∇0 ∇0 ψ − ψ) appears.
We want to show that ∇0 ∇0 ψ is negligible compared to the order ψ term in the Taylor’s
expansion of 8πG
c4
T00 . To do that it will be demonstrated that each term appearing in
(70) can be neglected.
ρmU
• First term of (70): This term is negligible since T00 ≈ ρMercury c2 and ρMercury
≈
10−30 .

• Second term of (70):

ȧ(t) H0 ∆ψ 1
3 ψ̇ ≈ 3 2 = 3 2 −1 ∆ψ . (71)
a(t) c0 ∆t c0 H0 ∆t

∆ψ represents the change of ψ in a time interval ∆t. Taking ∆t ≈ 4 × 109 years

(this is the time interval for which we have a bound for the variation of the radius
of Mercury), we get:
ȧ(t)
3 ψ̇ ≈ 10−46 ∆ψ cm−2 . (72)
a(t)
This quantity must be compared with:
8πG G
(−4) 4
ρψ ≈ 102 × 2 × ρm × ψ ≈ 10−26 ψ cm−2 . (73)
c0 c0

We see that the second term of (70) is also negligible.

16
References
[1] P. A. M. Dirac. The cosmological constants. Nature, 139:323, 1937.

[2] P. A. M. Dirac. A new basis for cosmology. Proc. Roy. Soc. London Series A,
165:199–208, 1938.

[3] J. K. Webb, V. V. Flambaum, C. W. Churchill, M. J. Drinkwater, and J. D.

Barrow. Search for time variation of the fine structure constant. Phys. Rev. Lett.,
83:884–887, 1999.

[4] J. K. Webb nd M. T. Murphy, V. V. Flambaum, V. A. Dzuba, J. D. Barrow, C. W.

Churchill, J. X. Prochaska, and A. M. Wolfe. Further evidence for cosmological
evolution of the fine structure constant. Phys. Rev. Lett., 87:091301, 2001.

[5] M. T. Murphy, J. K. Webb, V. V. Flambaum, V. A. Dzuba, C. W. Churchill, J. X.

Prochaska, J. D. Barrow, and A. M. Wolfe. Possible evidence for a variable fine-
structure constant from qso absorption lines: motivations, analysis and results.
Mon. Not. R. Astr. Soc., 327:1208–1222, 2001.

[6] M. T. Murphy, J. K. Webb, V. V. Flambaum, M. J. Drinkwater, F. Combes,

and T. Wiklind. Improved constraints on possible variation of physical constants
from h i 21-cm and molecular qso absorption lines. Mon. Not. R. Astr. Soc.,
327:1244–1248, 2001.

[7] M. T. Murphy, J. K. Webb, and V. V. Flambaum. Further evidence for a variable

fine-structure constant from keck/hires qso absorption spectra. Mon. Not. R. Astr.
Soc., 345:609–638, 2003.

[8] S. A. Levshakov, P. Molaro, S. Lopez, S. D’Odorico, M. Centurión, P. Bonifacio,

I. I. Agafonova, and D. Reimers. A new measure of ∆α/α at redshift z = 1.84
from very high resolution spectra of q1101-264. Astron. Astroph., 466:1077–1082,
2007.

[9] A. F. Martı́nez Fiorenzano, G. Vladilo, and P. Bonifacio. Search for alpha variation
in uves spectra: Analysis of c iv and si iv doublets towards qso 1101-264. Memorie
della Societa Astronomica Italiana Supplement, 3:252, 2003.

[10] R. Quast, D. Reimers, and S. A. Levshakov. Probing the variability of the fine-
structure constant with the vlt/uves. Astron. Astroph., 415:L7–L11, 2004.

[11] J. N. Bahcall, C. L. Steinhardt, and D. Schlegel. Does the fine-structure constant

vary with cosmological epoch? Astroph. J., 600:520–543, 2004.

[12] R. Srianand, H. Chand, P. Petitjean, and B. Aracil. Limits on the time variation of
the electromagnetic fine-structure constant in the low energy limit from absorption
lines in the spectra of distant quasars. Phys. Rev. Lett., 92:121302, 2004.

17
[13] J. W. Moffat. Superluminary universe: a possible solution to the initial value
problem in cosmology. Int. J. Mod. Phys. D, 2:351–365, 1993.

[14] A. Albrecht and J. Magueijo. Time varying speed of light as a solution to cosmo-
logical puzzles. Phys. Rev. D, 59:053516, 1999.

[15] J. D. Barrow. Cosmologies with varying light speed. Phys. Rev. D, 59:043515,
1999.

[16] J. Magueijo. Covariant and locally lorentz-invariant varying speed of light theories.
Phys. Rev. D, 62:103521, 2000.

[17] V. M. Canuto and S.-H. Hsieh. Scale covariance and g-varying cosmology. ii -
thermodynamics, radiation, and the 3 k background. Astroph. J. Supp., 41:243–
262, 1979.

[18] A. H. Taub. General relativistic variational principle for perfect fluids. Phys. Rev.,
94:1468–1470, 1954.

[19] B. F. Schutz. Perfect fluids in general relativity: Velocity potentials and a varia-
tional principle. Phys. Rev. D, 2:2762–2773, 1970.

[20] S. Chandrasekhar. An introduction to the study of stellar structure. Dover Publi-

cations, New York, 1957.

[21] J. P. Uzan. The fundamental constants and their variation: observational and
theoretical status. Rev. Mod. Phys., 75:403–455, 2003.

[22] M. W. McElhinny, S. R. Taylor, and D. J. Stevenson. Limits to the expansion

of earth, moon, mars and mercury and to changes in the gravitational constant.
Nature, 271:316–321, 1978.

[23] R. W. Siegfried and S. C. Solomon. Mercury - internal structure and thermal

evolution. Icarus, 23:192–205, 1974.

[24] Y Sortais, S Bize, M Abgrall, S. Zhang, Nicolas, C Mandache, Lemonde. P, P Lau-

rent, G Santarelli, N Dimarcq, P Petit, A. Clairon, A. Mann, A. Luiten, S. Chang,
and C. Salomon. Cold atom clocks. Physica Scripta, T95:50–57, 2000.

[25] R. Stothers. White dwarfs, the Galaxy and Dirac’s cosmology. Nature, 262:477–
479, August 1976.

[26] V. N. Mansfield and S. Malin. Astrophysical tests of scale-covariant theories.

Astroph. J., 237:349, 1980.

18