Logunov. The Theory of Gravity PDF

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1. The geometry of space-time . . . . . . . . . . . . . . . . . 22
2. The energy-momentum tensor of matter as the source of
the gravitational field . . . . . . . . . . . . . . . . . . . . 27
3. The gauge group of transformations . . . . . . . . . . . . 38
4. Density of the Lagrangian and the equations of motion
for the gravitational field proper . . . . . . . . . . . . . . 43
5. Equations of motion for the gravitational field and for
matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
6. The causality principle in RTG . . . . . . . . . . . . . . . 61
7. Mach’s principle . . . . . . . . . . . . . . . . . . . . . . . 69
8. Post-Newtonian approximation . . . . . . . . . . . . . . . 79
9. On the equality of inert and gravitational masses . . . . . 98
10. Evolution of the homogeneous and isotropic Universe . . 101
11. The gravitational field of a spherically symmetric static
body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Addendum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
12. Gravitational effects in the Solar system . . . . . . . . . . 154
12.1. Deflection of light rays by the Sun . . . . . . . . . . 164
12.2. The delay of a radiosignal . . . . . . . . . . . . . . 167
12.3. The shift of a planet’s perihelion . . . . . . . . . . . 175
12.4. The precession of a gyroscope . . . . . . . . . . . . 186
12.5. The gravitational shift
of spectral lines . . . . . . . . . . . . . . . . . . . . 189
13. Some other physical conclusions of RTG . . . . . . . . . 191
Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . 197
Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . 200
Appendix B∗ . . . . . . . . . . . . . . . . . . . . . . . . . 204
Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . 208
Appendix D . . . . . . . . . . . . . . . . . . . . . . . . . 212
Appendix E . . . . . . . . . . . . . . . . . . . . . . . . . 214
14. Elements of tensor analysis and of Riemannian geometry 216
ADDENDUM . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
On the gravitational force . . . . . . . . . . . . . . . . . . 234
Is the metric field of a non-inertial reference system a
special case of the gravitational physical field? . . . . . . 236
On the covariant conservation law . . . . . . . . . . . . . 237
H.Poincare On the dynamics of the electron . . . . . . . 241
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Preface
This monograph sums up studies performed in developing the
relativistic theory of gravity (RTG) and presented in refs. [3, 9,
arXiv:gr-qc/0210005 v2 21 Oct 2002
38, 10, 5, 11, 6, 34, 12, 35, 36, 37, 31, 13]. Detailed references
to earlier works, that to a certain extent served as scaffolding
in the construction of RTG, are given in the monograph [10],
written together with prof. M. A. Mestvirishvili and published
in 1989. Therein, also, critical comments are presented con-
cerning general relativity theory (GRT), which still remain in
force. In order to facilitate reading in section 14 we provide
elements of tensor analysis and Riemannian geometry. As a
rule, we make use of the set of units in which G = c = h̄ = 1.
However, in the final expressions we restore the dependence
on the constants G, c, h̄. Throughout the book, Greek letters
assume values 0,1,2,3, while Latin letters assume —1,2,3.
The creation of this monograph advanced together with
the completion of studies of individual issues, so it inevitably
contains recurrences, especially concerning such issues that are
important for understanding the essence of both RTG and
GRT.
The hypothesis underlying RTG asserts that the gravita-
tional field, like all other physical fields, develops in Minkowski
space, while the source of this field is the conserved energy-
momentum tensor of matter, including the gravitational field
itself. This approach permits constructing, in a unique man-
ner, the theory of the gravitational field as a gauge theory.
Here, there arises an effective Riemannian space, which lit-
erally has a field nature. In GRT the space is considered to
be Riemannian owing to the presence of matter, so gravity is
considered a consequence of space–time exhibiting curvature.
The RTG gravitational field has spins 2 and 0 and represents a
physical field in the Faraday–Maxwell spirit. The complete set
of RTG equations follows directly from the least action prin-
ciple. Since all physical fields develop in Minkowski space, all
5
fundamental principles of physics — the integral conservation
laws of energy–momentum and of angular momentum — are
strictly obeyed in RTG. In the theory the Mach principle is
realized: an inertial system is determined by the distribution
of matter. Unlike GRT, acceleration has an absolute sense.
Inertial and gravitational forces are separated, and they differ
in their nature. The theory, unlike GRT, provides a unique
explanation for all gravitational effects in the Solar system.
GRT does not comply with the equivalence principle,
does not explain the equality of the inert and active
gravitational masses, and gives no unique prediction
for gravitational effects. It does not contain the usual
conservation laws of energy–momentum and of angu-
lar momentum of matter.
It should be especially noted that the known post-New-
tonian approximation do satisfy the equivalence principle, do
provide a unique description of gravitational effects in the So-
lar system, and also establish the equality between the inertial
and active gravitational masses. However, it does not follow
uniquely from the GRT equations, since its derivation relies on
additional assumptions, that do not follow from the theory, i.e.
a departure occurs beyond the limits of GRT, which is based
on the gravitational field being represented as a physical field,
although this is not so in GRT. Therefore, this approxima-
tion cannot be considered a unique consequence of the GRT
equations. It has rather been guessed, then derived from the
theory, while, according to RTG, the post–Newtonian approx-
imation follows uniquely from equations of the theory. Thus,
the post–Newtonian approximation, previously applied for the
description of gravitational effects follows directly from our
theory. RTG introduces essential changes into the character
of the development of the Universe and into the collapse of
large masses.
Analysis of the development of a homogeneous and isotropic
Universe within RTG leads to the conclusion that the Universe
is infinite, and that it is “flat”. Its development proceeds cycli-
6
cally from a certain maximum density down to a minimum and
so on. Thus, no pointlike Big Bang occurred in the past. There
existed a state of high density and high temperature at each
point in space.
According to RTG, the so-called cosmological “expansion”
of the Universe, observed by the red shift, is explained by
changes in the gravitational field, but not by relative motion
— galaxies escaping from each other, which actually does not
take place. Matter in the Universe is in a state of rest rela-
tive to an inertial coordinate system. The peculiar velocities
of galaxies relative to an inertial system arose owing to a cer-
tain structure of the inhomogeneity of the distribution of mat-
ter during the period, when the Universe became transparent.
This means that in the past the distance between galaxies was
never zero. The theory predicts the existence in the Universe
of a large hidden mass of “dark matter”. According to RGT,
“black holes” cannot exist: a collapsing star cannot disappear
beyond its gravitational radius. Objects with large masses can
exist, and they are characterized not only by mass, but also
by a distribution of matter density. Since, in accordance with
GRT, objects with masses exceeding three solar masses trans-
form, at the conclusive stage of their evolution, into “black
holes”, an object found to have a large mass is usually at-
tributed to “black holes”. Since RTG predictions concerning
the behaviour of large masses differ essentially from GRT pre-
dictions, observational data of greater detail are required for
testing the conclusions of theory. Thus, for example, in RTG
spherically symmetric accretion of matter onto a body of large
mass, that is at its conclusive stage of evolution (when the nu-
clear resources are exhausted), will be accompanied by a sig-
nificant release of energy owing to the fall of matter onto the
body’s surface, while in GRT the energy release in the case of
spherically symmetric accretion of matter onto a “black hole”
is extremely small, since the falling matter takes the energy
with it into the “black hole”. Observational data on such ob-
jects could answer the question whether “black holes” exist in
7
Nature, or not. Field concepts of gravity necessarily require in-
troduction of the graviton rest mass, which can be determined
from observational data: the Hubble “constant” and the decel-
eration parameter q. According to the theory, the parameter
q can only be positive, at present, i.e. deceleration of “ex-
pansion” of the Universe takes place, instead of acceleration.
For this reason, the latest observational data on acceleration
of the “expansion” must be checked carefully, since the con-
clusions of theory concerning “deceleration” follow from the
general physical principles mentioned above.
I sincerely wish to thank my teacher acad. N.N. Bogolubov
who, during the hard years of searches and struggle, provided
spiritual support as well as valuable advice that stimulated the
research.
I am grateful to Providence for my wife Anna Nikolaevna,
who for over forty years was my support.
I am profoundly grateful to prof. M.A. Mestvirishvili for
many years of joint work on the construction of relativistic
theory of gravity.
I am grateful to acad. A.M. Baldin, acad. V.S. Vladimirov,
acad. V.G. Kadyshevsky, acad. A.N. Tavkhelidze for valuable
discussions.
I take advantage of the occasion to express my deep grati-
tude to professors S.S. Gershtein, V.I. Denisov, Yu.M. Losku-
tov, to associate professor A.A. Vlasov and Candidate of phy-
sics and mathematics Yu.V. Chugreev for common work and
for numerous discussions of the problems at issue. I am also
grateful to professors V.A. Petrov, N.E.Tyurin, A.A. Tyapkin
and O.A. Khrustalev for useful discussions.
I express profound gratitude to acad. A.M. Baldin, corre-
sponding member of RAS S.S. Gershtein and prof. M.A. Mest-
virishvili, who read the entire manuscript and made a number
of valuable advices and comments.
A.A. Logunov
April 2000
Introduction
Since construction of the relativistic theory of gravity (RTG) is
based on special relativity theory (SRT), we shall deal with the
latter in greater detail and in doing so we shall examine both
the approach of Henri Poincaré and that of Albert Einstein.
Such an analysis will permit a more profound comprehension
of the difference between these approaches and will make it
possible to formulate the essence of relativity theory.
In analyzing the Lorentz transformations, H. Poincaré sho-
wed that these transformations, together with all spatial ro-
tations, form a group that does not alter the equations of
electrodynamics. Richard Feynman wrote the following about
this: “Precisely Poincaré proposed investigating what could be
done with the equations without altering their form. It was
precisely his idea to pay attention to the symmetry properties
of the laws of physics” 1. H.Poincaré did not restrict himself to
studying electrodynamics; he discovered the equations of rel-
ativistic mechanics and extended the Lorentz transformations
to all the forces of Nature. Discovery of the group, termed
by H.Poincaré the Lorentz group, made it possible for him
to introduce four-dimensional space-time with an invariant
subsequently termed the interval
dσ 2 = (dX 0 )2 − (dX 1 )2 − (dX 2 )2 − (dX 3 )2 . (α)

Precisely from the above it is absolutely clear that time and
spatial length are relative.
Later, a further development in this direction was made by
Herman Minkowski, who introduced the concepts of timelike
and spacelike intervals. Following H.Poincaré and H.Minkow-
ski exactly, the essence of relativity theory may be formulated
thus: all physical phenomena proceed in space–time,
the geometry of which is pseudo-Euclidean and is de-
termined by the interval (α). Here it is important to em-
1
R.Feynman. The character of physical laws. M.:Mir, 1968, p.97.
9
phasize, that the geometry of space-time reflects those
general dynamic properties, that represent just what
makes it universal. In four-dimensional space (Minkowski
space) one can adopt a quite arbitrary reference frame
X ν = f ν (xµ ) ,
realizing a mutually unambiguous correspondence with a Ja-
cobian differing from zero. Determining the differentials
∂f ν µ
dX ν = dx ,
∂xµ
and substituting these expressions into (α) we find
dσ 2 = γµν (x)dxµ dxν , where
∂f σ ∂f σ
γµν (x) = %σ , %σ = (1, −1, −1, −1) . (β)
∂xµ ∂xν
It is quite evident that the transition undergone to an ar-
bitrary reference system did not lead us beyond the limits of
pseudo-Euclidean geometry. But hence it follows that non-
inertial reference systems can also be applied in SRT. The
forces of inertia arising in transition to an accelerated refer-
ence system are expressed in terms of the Christoffel symbols of
Minkowski space. The representation of SRT stemming from
the work of H.Poincaré and H.Minkowski was more general
and turned out to be extremely necessary for the construction
of SRT, since it permitted introduction of the metric tensor
γµν (x) of Minkowski space in arbitrary coordinates and thus
made it possible to introduce in a covariant manner the grav-
itational field, upon separation of the forces of inertia from
gravity.
From the point of view of history it must be noted that
in his earlier works 2 , “The measurement of time” and “The
2
H.Poincaré. The principle of relativity. M.:Atomizdat, 1973, pp.19,
33.
10
present and future of mathematical physics”, H.Poincaré dis-
cussed in detail issues of the constancy of the velocity of light,
of the simultaneity of events at different points of space deter-
mined by the synchronization of clocks with the aid of a light
signal. Later, on the basis of the relativity principle, which he
formulated in 1904 for all physical phenomena, as well as on
the work published by H.Lorentz the same year, H.Poincaré
discovered a transformation group in 1905 and termed
it the Lorentz group. This permitted him to give the
following essentially accurate formulation of the rela-
tivity theory: the equations of physical processes must
be invariant relative to the Lorentz group. Precisely
such a formulation was given by A.Einstein in 1948: “With
the aid of the Lorentz transformations the special principle of
relativity can be formulated as follows: The laws of Nature are
invariant relative to the Lorentz transformation (i.e. a law
of Nature should not change if it is referred to a new inertial
reference frame with the aid of the Lorentz transformation for
x, y, z, t)” 3 .
The existence of a group of coordinate-time transforma-
tions signifies that there exists an infinite set of equivalent
(inertial) reference frames related by the Lorentz transforma-
tions. From the invariance of equations it follows, in a trivial
manner, that physical equations in the reference frames x and
x! , related by the Lorentz transformations, are identical. But
this means that any phenomenon described both in x and x!
reference systems under identical conditions will yield identi-
cal results, i.e. the relativity principle is satisfied in a triv-
ial manner. Certain, even prominent, physicists understood
this with difficulty not even long ago, while others have not
even been able to. There is nothing strange in this fact, since
any study requires certain professionalism. What is surprising
is the following: they attempt to explain their incomprehen-
3
Einstein A. Collection of scientific works, Moscow: Nauka, 1966,
vol.2, art.133, p.660.
11
sion, or the difficulty they encountered in understanding, by
H.Poincaré allegedly “not having taken the decisive step”, “not
having gone to the end”. But these judgements, instead of the
level of the outstanding results achieved by H.Poincaré in rel-
ativity theory, characterize their own level of comprehension
of the problem.
Precisely for this reason W.Pauli wrote the following in
1955 in connection with the 50-th anniversary of relativity
theory: “Both Einstein and Poincaré relied on the preparatory
works performed by H.A.Lorentz, who was very close to the
final result, but was not able to take the last decisive step. In
the results, obtained by Einstein and Poincaré independently
of each other, being identical I see the profound meaning of the
harmony in the mathematical method and analysis performed
with the aid of thought experiments and based on the entire set
of data of physical experiments” 4.
Detailed investigation by H.Poincaré of the Lorentz group
invariants resulted in his discovery of the pseudo-Euclidean
geometry of space-time. Precisely on such a basis, he estab-
lished the four-dimensionality of physical quantities: force, ve-
locity, momentum, current. H.Poincaré’s first short work ap-
peared in the reports of the French Academy of sciences before
A.Einstein’s work was even submitted for publication. That
work contained an accurate and rigorous solution of the prob-
lem of electrodynamics of moving bodies, and at the same time
it extended the Lorentz transformations to all natural forces,
of whatever origin they might be. Very often many historians,
and, by the way, physicists, also, discuss priority issues. A very
good judgement concerning this issue is due to academicians
V.L.Ginzburg and Ya.B.Zel’dovich, who in 1967 wrote:“Thus,
no matter what a person has done himself, he cannot claim
4
W.Pauli. Essays in physics. M.:Nauka, 1975, p.189.
12
priority, if it later becomes known that the same result was
obtained earlier by others” 5.
A.Einstein proceeded toward relativity theory from an anal-
ysis of the concepts of simultaneity and of synchronization for
clocks at different points in space on the basis of the principle
of constancy of the velocity of light. ¡¡Each ray of light travels
in a reference frame at “rest” with a certain velocity V , inde-
pendently of whether this ray of light is emitted by a body at
rest or by a moving body.¿¿ But this point cannot be consid-
ered a principle, since it implies a certain choice of reference
frame, while a physical principle should clearly not depend
on the method of choosing the reference frame. In essence,
A.Einstein accurately followed the early works of H.Poincaré.
However, within such an approach it is impossible to arrive at
non-inertial reference frames, since in such reference frames it
is impossible to take advantage of clock synchronization, so the
notion of simultaneity loses sense, and, moreover, the velocity
of light cannot be considered constant.
In a reference frame undergoing acceleration the proper
time dτ , where
γ0α dxα γ0i γ0k
dσ 2 = dτ 2 − sik dxi dxk , dτ = √ , sik = −γik +
γ00 γ00
is not a complete differential, so the synchronization of clocks

at different points in space depends on the synchronization
path. This means that such a concept cannot be applied for
reference frames undergoing acceleration. It must be stressed
that the coordinates in expression (β) have no metric meaning,
on their own. Physically measurable quantities must be con-
structed with the aid of coordinates and the metric coefficients
γµν . But all this remained misunderstood for a long time in
SRT, since it was usual to adopt A.Einstein’s approach, in-
stead of the one of H.Poincaré and H.Minkowski. Thus, the
5
V.L.Ginzburg, Ya.B.Zel’dovich. Familiar and unfamiliar Zel’dovich.
M.:Nauka, 1993, p.88.
13
starting points introduced by A.Einstein were of an exclusively
limited and partial nature, even though they could create an
illusion of simplicity. It was precisely for this reason that even
in 1913 A.Einstein wrote: “In usual relativity theory only lin-
ear orthogonal transformations are permitted” 6 . Or somewhat
later, in the same year, he writes: “In the original relativity
theory the independence of physical equations of the specific
choice of reference system is based on postulating the funda-
mental invariant ds2 =
! 2
dxi , while now the issue consists
in constructing a theory (general relativity theory is implied –
A.L.), in which the role of the fundamental invariant is per-
formed by a linear element of the general form
ds2 = gik dxi dxk ” 7 .

"
i,k
A.Einstein wrote something similar in 1930: “In special rel-

ativity theory only such coordinate changes (transformations)
are allowed that provide for the quantity ds2 (a fundamental
invariant) in the new coordinates having the form of the sum
of square differentials of the new coordinates. Such transfor-
mations are called Lorentz transformations” 8.
Hence it is seen that the approach adopted by A.Einstein
did not lead him to the notion of space-time exhibiting a
pseudo-Euclidean geometry. A comparison of the approaches
of H.Poincaré and A.Einstein to the construction of SRT clearly
reveals H.Poincaré’s approach to be more profound and gen-
eral, since precisely H.Poincaré had defined the pseudo-Eu-
clidean structure of space-time. A.Einstein’s approach essen-
tially restricted the boundaries of SRT, but, since the exposi-
tion of SRT in the literature usually followed A.Einstein, SRT
6
Einstein A. Collection of scientific works. Moscow: Nauka, 1965,
vol.1, art.21, p.232.
7
vol.1, art.22,p.269.
8
vol.2, art.95, p.281.
14
was quite a long time considered valid only in inertial reference
systems. Minkowski space was then treated like a useful geo-
metric interpretation or like a mathematical formulation of the
principles of SRT within the approach of Einstein. Let us now
pass over to gravity. In 1905 H.Poincaré wrote: “... that forces
of whatever origin, for example, the forces of gravity, behave
in the case of uniform motion (or, if you wish, under Lorentz
transformations) precisely like electromagnetic forces” 9. This
is precisely the path we shall follow.
A.Einstein, having noticed the equality of inertial and grav-
itational masses, was convinced that the forces of inertia and
of gravity are related, since their action is independent of a
body’s mass. In 1913 he arrived at the conclusion that, if in
expression (α) “... we introduce new coordinates x1 , x2 , x3 , x4 ,
with the aid of some arbitrary substitution, then the motion
of a point relative to the new reference frame will proceed in
accordance with the equation
#
δ{ ds} = 0 ,
and
ds2 = gµν dxµ dxν .”
"
µ,ν
and he further pointed out: ¡¡The motion of a material point in

the new reference system is determined by the quantities gµν ,
which in accordance with the preceding paragraphs should be
understood as the components of the gravitational field, as soon
as we decide to consider this new system to be “at rest”¿¿ 10 .
Identifying in such a manner the metric field, obtained
from (α) with the aid of coordinate transformations,
and the gravitational field is without physical grounds,
since transformations of coordinates do not lead us
beyond the framework of pseudo-Euclidean geometry.
9
H.Poincaré. Special relativity principle. M.:Atomizdat, 1973, p.152.
10
vol.1, art.23, p.286.
15
From our point of view, it is not permitted to consider such
a metric field as the gravitational field, since this contradicts
the very essence of the concept of a field as a physical reality.
Therefore, it is impossible to agree with the following reasoning
of A.Einstein: ¡¡The gravitational field “exists” with respect to
the system K ! in the same sense as any other physical quantity
that can be defined in a certain reference system, even though
it does not exist in system K. There is nothing strange, here,
and it may be readily demonstrated by the following example
taken from classical mechanics. Nobody doubts the “reality”
of kinetic energy, since otherwise it would be necessary to re-
nounce energy in general. It is clear, however, that the kinetic
energy of bodies depends on the state of motion of the reference
system: by an appropriate choice of the latter it is evidently
possible to provide for the kinetic energy of uniform motion
of a certain body to assume, at a certain moment of time, a
given positive or zero value set beforehand. In the special case,
when all the masses have equal in value and equally oriented
velocities, it is possible by an appropriate choice of the refer-
ence system to make the total kinetic energy equal to zero. In
my opinion the analogy is complete¿¿ 11.
As we see, Einstein renounced the concept of a classical
field, such as the Faraday–Maxwell field possessing density of
energy-momentum, in relation to the gravitational field. Pre-
cisely this path led him up to the construction of GRT, to
gravitational energy not being localizable, to introduction of
the pseudotensor of the gravitational field. If the gravitational
field is considered as a physical field, then it, like all other
physical fields, is characterized by the energy-momentum ten-
sor tµν . If in some reference frame, for instance, K ! , there ex-
ists a gravitational field, this means that certain components
(or all of them) of the tensor tµν differ from zero. The tensor
tµν cannot be reduced to zero by a coordinate transformation,
11
vol.1, art.46, p.620.
16
i..e, if a gravitational field exists, then it represents a physical
reality, and it cannot be annihilated by a choice of reference
system. It is not correct to compare such a gravitational field
with kinetic energy, since the latter is not characterized by a
covariant quantity. It must be noted that such a comparison
is not admissible, also, in GRT, since the gravitational field in
this theory is characterized by the Riemann curvature tensor.
If it differs from zero, then the gravitational field exists, and
it cannot be annihilated by a choice of reference system, even
locally.
Accelerated reference systems have played an important
heuristic role in A.Einstein’s creative work, although they have
nothing to do with the essence of GRT. By identifying accel-
erated reference systems to the gravitational field, A.Einstein
came to perceive the metric space-time tensor as the principal
characteristic of the gravitational field. But the metric ten-
sor reflects both the natural properties of geometry and the
choice of reference system. In this way the possibility arises of
explaining the force of gravity kinematically, by reducing it to
the force of inertia. But in this case it is necessary to renounce
the gravitational field as a physical field. “Gravitational fields
(as A.Einstein wrote in 1918) may be set without introducing
tensions and energy density.” 12 . But that is a serious loss,
and one cannot consent to it. However, as we shall further
see, this loss can be avoided in constructing RTG.
Surprisingly, even in 1933 A.Einstein wrote: ¡¡In special
Relativity theory — as shown by H.Minkowski — this metric
was quasi-Euclidean, i.e. the square “length” ds of a linear ele-
ment represented a certain quadratic function of the coordinate
differentials. If, on the other hand, new coordinates are intro-
duced with the aid of a linear transformation, then ds2 remains
a homogeneous function of the coordinate differentials, but the
coefficients of this function (gµν ) will no longer be constant,
12
vol.1, art.47, p.627.
17
but certain functions of the coordinates. From a mathematical
point of view this means that the physical (four-dimensional)
space possesses a Riemannian metric¿¿ 13 .
This is certainly wrong, since a pseudo-Euclidean metric
cannot be transformed into a Riemannian metric by transfor-
mation of the coordinates. But the main point, here, con-
sists in something else, namely, in that in this way, thanks to
his profound intuition, A.Einstein arrived at the necessity of
introducing precisely Riemannian space, since he considered
the metric tensor gµν of this space to describe gravity. This
was essentially how the tensor nature of gravity was revealed.
The unity of the Riemannian metric and gravity is the main
principle underlying general relativity theory. V.A.Fock wrote
about this principle: “... precisely this principle represents
the essence of Einstein’s theory of gravity” 14 . From a general
point of view, however, the answer to the following question
still remains unclear: why is it necessary to relate gravity pre-
cisely to Riemannian space, and not to any other.
The introduction of Riemannian space permitted using the
scalar curvature R as the Lagrangian function and, with the
aid of the least action principle, to obtain the Hilbert–Einstein
equation. Thus, the construction of Einstein’s general relativ-
ity theory was completed. Here, as particularly stressed by
J.L.Synge: “In Einstein’s theory the presence or absence of a
gravitational field depends on whether the Riemann tensor dif-
fers from or equals to zero. This is an absolute property, which
is in no way related to the world line of any observer” 15.
In GRT, however, difficulties arose with the conservation
laws of energy-momentum and angular momentum. D.Hilbert
wrote in this connection: “... I claim that within general rel-
13
vol.2, art.110, p.405.
14
V.A.Fock. Theory of space, time and gravity. M.:Gostekhizdat,
1961, p.308.
15
J.L.Synge. Relativity: the general theory. M.:Foreign literature
publishers, 1963, p.9.
18
ativity theory, i.e. in the case of general invariance of the
Hamiltonian function, there definitely exist no energy equa-
tions ... corresponding to the energy equations in orthogonal-
invariant theories, I could even point to this circumstance as
a characteristic feature of general relativity theory” 16 . All the
above is explained by the absence in Riemannian space of the
ten-parameter group of motion of space-time, so it is essen-
tially impossible to introduce energy-momentum and angular
momentum conservation laws, similar to those that hold valid
in any other physical theory.
Another feature peculiar to GRT, as compared to known
theories, consists in the presence of second-order derivatives
in the Lagrangian function R. About fifty years ago Nathan
Rosen demonstrated that if, together with the Riemannian
metric gµν one introduces the metric γµν of Minkowski space,
then it becomes possible to construct the scalar density of the
Lagrangian of the gravitational field, which will not contain
derivatives of orders higher than one. Thus, for example, he
constructed such a density of the Lagrangian which led to the
Hilbert-Einstein equations. Thus came into being the bimetric
formalism. However, such an approach immediately compli-
cated the problem of constructing a theory of gravity, since,
when using the tensors gµν and γµν , one can write out a large
number of scalar densities, and it is absolutely not clear which
scalar density must be chosen as the Lagrangian density for
constructing the theory of gravity. Although the GRT math-
ematical apparatus does permit introducing, instead of ordi-
nary derivatives, covariant derivatives of Minkowski space, the
metric γµν not being present in the Hilbert-Einstein equations
renders its utilization in GRT devoid of any physical meaning,
because the solutions for the metric gµν are independent of the
choice of γµν . It must be noted that substitution of covariant
derivatives for ordinary derivatives in Minkowski space leaves
the Hilbert-Einstein equations intact. This is explained by
16
V.P.Vizgin. Relativistic theory of gravity. M.:Nauka, 1981, p.319.
19
the fact that, if in Minkowski space one substitutes covariant
derivatives for ordinary ones in the Riemann curvature tensor,
it will not change. Such a substitution in the Riemann tensor
is nothing, but an identical transformation. Precisely for this
reason such a freedom in writing out the Riemann tensor can-
not be taken as an advantage within the framework of GRT,
since the metric tensor of Minkowski space does not enter into
the Hilbert-Einstein equations.
In constructing RTG, this freedom in writing the Riemann
tensor turns out to be extremely necessary. But in this case
the metric of Minkowski space enters into the equations of
the gravitational field, and the field itself is considered as a
physical field in Minkowski space. In GRT we only deal with
the metric of Riemannian space as the main characteristic of
gravity, in which both the features of geometry itself and the
choice of reference frame are reflected. When the gravitational
interaction is switched off, i.e. when the Riemann curvature
tensor equals zero, we arrive at Minkowski space. It is precisely
for this reason that in GRT the problem arises of satisfying
the equivalence principle, since it is impossible to determine
in which reference frame (inertial or accelerated) we happened
to be when the gravitational field was switched off.
The relativistic theory of gravity, presented in this work is
constructed as a field theory of the gravitational field within
the framework of special relativity theory. The starting point
is the hypothesis that the energy-momentum tensor — which
is a universal characteristic of matter — serves as the source of
gravity. The gravitational field is considered to be a universal
physical field with spins 2 and 0, owing to the action of which
the effective Riemannian space arises. This permits to find the
gauge group and to construct unambiguously the Lagrangian
density of the gravitational field. The set of equations of this
theory is generally covariant and form-invariant with respect
to the Lorentz group. Here, it is necessary in the theory to
introduce the graviton mass. The graviton mass essentially in-
20
fluences the evolution of the Universe and alters the character
of the gravitational collapse.
The goal of this work is a further development of the ideas
by H.Poincaré, H.Minkowski, A.Einstein, D.Hilbert, N.Rosen,
V.A.Fock, S.Gupta, V.Thirring, R.Feynman, S.Weinberg in
the domain of theory of relativity and gravity.
1. The geometry of space-time
In Chapter II, “Space and time”, of his book “Recent ideas”,
H.Poincaré wrote: “The principle of physical relativity may
serve for defining of space. It can be said to provide us with
a novel instrument for measurement. I shall explain. How
can a solid body serve for measuring or, to be more correct,
for constructing space? The point is the following: in trans-
ferring a solid body from one place to another we, thus, note
that it can be first applied to one figure, then to another, and
we conventionally agree to consider these figures equal to each
other. Geometry originated from this convention. Geometry is
nothing but a science of mutual interrelationships between such
transformations or, speaking in the mathematical language, a
science of the structure of the group formed by these transfor-
mations, i.e. of the group of motions of solid bodies.
Now, consider another group, the group of transformations
not altering our differential equations. We arrive at a new way
for defining the equality between two figures. We no longer
say: two figures are equal, if one and the same solid body can
be applied to both one and the other figures. We shall say: two
figures are equal, when one and the same mechanical system,
so distant from its neighbours that it may be considered iso-
lated, being first thus situated so its material points reproduce
the first figure, and then so they reproduce the second figure,
behaves in the second case precisely like in the first. Do these
two approaches differ in essence? No.
A solid body represents a mechanical system, just like any
other. The only difference between the previous and the new
definitions of space consists in that the latter is broader, since
it allows substitution of any mechanical system for the solid
body. Moreover, our new convention not only defines space,
but time, also. It provides us with an explanation of what are
22
two equal time intervals or of what is represented by a time
interval twice as long as another” 17 .
Precisely in this way, by discovering the group of transfor-
mations not altering the Maxwell–Lorentz equations, H.Poin-
caré introduced the notion of four-dimensional space-time ex-
hibiting pseudo-Euclidean geometry. This concept of geometry
was later developed by H.Minkowski.
We have chosen the pseudo-Euclidean geometry of space-
time as the basis of the relativistic theory of gravity presently
under development, since it is the fundamental Minkowski
space for all physical fields, including the gravitational field.
Minkowski space cannot be considered to exist a priori, since
it reflects the properties of matter and, hence, cannot be sep-
arated from it. Although formally, precisely owing to the
structure of space being independent of the form of matter,
it is sometimes dealt with abstractly, separately from mat-
ter. In Galilean coordinates of an inertial reference system in
Minkowski space, the interval that characterizes the structure
of geometry and that is an invariant by construction, has the
form
dσ 2 = (dx0 )2 − (dx1 )2 − (dx2 )2 − (dx3 )2 .
Here dxν represent differentials of the coordinates. In spite of
the fact that the interval dσ, as a geometric characteristic of
space-time, is independent of the choice of reference system,
which is due to its very construction, one can still encounter
in modern text-books on theoretical physics (see, for instance,
Ref. [4]) “proofs” of the interval being the same in all inertial
reference systems although it is an invariant and is indepen-
dent of the choice of reference system.
Even such an outstanding physicist as L.I.Mandelstam
wrote in his book [17]: “... special relativity theory cannot an-
swer the question, how a clock behaves when moving with ac-
celeration and why it slows down, because it does not deal with
reference systems moving with acceleration”. The incorrect as-
17
H.Poincaré. On science. M.:Nauka, 1938, p.427.
23
sertions in [27, 19, 20, 30] can be explained by Minkowski space
being considered by many people to be only some formal geo-
metrical interpretation of SRT within A.Einstein’s approach,
instead of a revelation of the geometry of space-time. The
issues of such limited concepts as the constancy of the speed
of light, the synchronization of clocks, the speed of light be-
ing independent of the motion of its source became the most
discussed topics. All this narrowed the scope of SRT and re-
tarded the understanding of its essence. And its essence
actually consists only in that the geometry of space-
time, in which all physical processes occur, is pseudo-
Euclidean.
In an arbitrary reference system the interval assumes the
form
dσ 2 = γµν (x)dxµ dxν ,
γµν (x) is the metric tensor of Minkowski space. We note that
one cannot, in principle, speak of the synchronization of clocks
or of the constancy of the speed of light in an non-inertial
reference system [7]. Most likely, precisely the lacking clarity
on the essence of SRT led A.Einstein to concluding: “that
within the framework of special relativity theory there is no
place for a satisfactory theory of gravity” 18 . Free motion of a
test body in an arbitrary reference system takes place along a
geodesic line of Minkowski space:
DU ν dU ν ν
= + γαβ U αU β = 0 ,
dσ dσ
dxν ν
where U ν = dσ , γαβ (x) are Christoffel symbols defined by
the expression
ν 1
γαβ (x) = γ νσ (∂α γβσ + ∂β γασ − ∂σ γαβ ) .
2
18
vol.4, art.76, p.282.
24
In 1921, in the article “Geometry and experiment”, A. Ein-
stein wrote: “The issue of whether this continuum has an Eu-
clidean, Riemannian or any other structure is a physical is-
sue, which can only be settled by experiment, and not an issue
of convention concerning a choice of simple expedience...” 19.
This is, naturally, correct. But there immediately arises a
question: what experiment? There may exist quite many ex-
perimental facts. Thus, for example, it is possible, in principle,
by studying the motion of light and of test bodies, to establish
unambiguously the geometry of space-time. Must a physical
theory be based on it? At first sight, the answer to this ques-
tion could be positive. And the issue would seem settled. Pre-
cisely such was the path that A.Einstein took in constructing
GRT. Test bodies and light move along geodesic lines of Rie-
mannian space-time. So he based the theory on Riemannian
space. However, the situation is much more complex. All types
of matter satisfy conservation laws of energy-momentum and
of angular momentum. Precisely these laws, that originated
from a generalization of numerous experimental data, charac-
terize the general dynamic properties of all forms of matter by
introducing universal characteristics permitting quantitative
description of the transformation of some forms of matter into
others. And all this also represents experimental facts, which
have become fundamental physical principles. What should
be done with them? If one follows A.Einstein and retains Rie-
mannian geometry as the basis, then they must be discarded.
That price would be too high. It is more natural to retain them
for all physical fields, including the gravitational field. But, in
this case, theory must, then, be based on Minkowski space,
i.e. on the pseudo-Euclidean geometry of space-time. We
have adopted precisely this approach, following H.Poincaré.
The fundamental principles of physics, that reflect the nu-
merous available experimental facts, indicate what geometry
19
vol.2, art.61, p.87.
25
of the space-time it is actually necessary to use as the ba-
sis of gravity theory. Thus, the issue of the structure of the
space-time geometry is actually a physical issue, that should
be resolved by experiment, and, from our point of view, the
structure of the geometry of space-time is not determined by
specific experimental data on the motion of test bodies and
of light, but by fundamental physical principles based on the
entire set of existing experimental facts. It is precisely here
that our initial premises for constructing the theory of gravity
differ completely from the ideas applied by A.Einstein as the
basis of GRT. But they are fully consistent with the ideas of
H.Poincaré.
We have chosen the pseudo-Euclidean geometry of space-
time as the basis of the relativistic theory of gravity, but that
certainly does not mean that the effective space will also be
pseudo-Euclidean. The influence of the gravitational field may
be expected to lead to a change in the effective space. We
shall deal with this issue in detail in the next section. The
metric of Minkowski space permits introducing the concepts
of standard length and time intervals, when no gravitational
field is present.
2. The energy-momentum tensor of
matter as the source of the
gravitational field
Owing to the existence in Minkowski space of the Poincaré ten-
parameter group of motion, there exist for any closed physical
system ten integrals of motion, i.e. the conservation laws of
energy-momentum and angular momentum hold valid. Any
physical field in Minkowski space is characterized by the den-
sity of the energy-momentum tensor tµν , which is a general
universal characteristic of all forms of matter that satisfies
both local and integral conservation laws. In an arbitrary ref-
erence system the local conservation law is written in the form
Dµ tµν = ∂µ tµν + γαβ

ν αβ
t =0.
Here tµν is the total conserved density of the energy-momentum

tensor for all the fields of matter; Dµ represents the covariant
derivative in Minkowski space. Here and further we shall al-
ways deal with the densities of scalar and tensor quantities
defined in accordance with the rule
√ √
φ̃ = −γφ , φ̃µν = −γφµν , γ = det(γµν ) .
The introduction of densities is due to an invariant volume el-

ement in Minkowski space being determined by the expression
√
−γd4 x ,
while an invariant volume element in Riemannian space is

given by the expression
√
−gd4 x , g = det(gµν ) .
Therefore, the principle of least action assumes the form

#
δS = δ Ld4 x = 0 ,
27
where L is the scalar density of the Lagrangian of matter. In
deriving Euler’s equations with the aid of the principle of least
action we shall automatically have to deal precisely with the
variation of the Lagrangian density. According to D.Hilbert,
the density of the energy-momentum tensor tµν is expressed
via the scalar density of the Lagrangian L as follows:
δL
tµν = −2 , (2.1)
δγµν
where
$ %
δL ∂L ∂L ∂γµν
= − ∂σ , γµν,σ = .
δγµν ∂γµν ∂γµν,σ ∂xσ
Owing to gravity being universal, it would be nat-

ural to assume the conserved density of the energy-
momentum tensor of all fields of matter, tµν , to be
the source of the gravitational field. Further, we shall
take advantage of the analogy with electrodynamics, in which
the conserved density of the charged vector current serves as
the source of the electromagnetic field, while the field itself is
described by the density of the vector potential Ãν :
Ãν = (φ̃, Ã) .
In the absence of gravity, Maxwell’s equations of electrody-

namics will have the following form in arbitrary coordinates:
γ αβ Dα Dβ Ãν + µ2 Ãν = 4πj ν ,

D ν Ãν = 0 ,
Here, for generalization we have introduced the parameter µ,

which, in the system of units h̄ = c = 1 is the photon rest
mass.
Since we have decided to consider the conserved density of
the energy-momentum tµν to be the source of the gravitational
field, it is natural to consider the gravitational field a tensor
28
field and to describe it by the density of the symmetric tensor
φ̃µν : √
φ̃µν = −γφµν ,
and in complete analogy with Maxwell’s electrodynamics the
equations for the gravitational field can be written in the form
γ αβ Dα Dβ φ̃µν + m2 φ̃µν = λtµν , (2.2)

Dµ φ̃µν = 0 . (2.3)
Here λ is a certain constant which, in accordance with the

principle of correspondence to Newton’s law of gravity, should
be equal to 16π. Equation (2.3) excludes spins 1 and 0! , only
retaining those polarizational properties of the field, that cor-
respond to spins 2 and 0.
The density of the energy-momentum tensor of matter tµν
consists of the density of the energy-momentum tensor of the
gravitational field, tµν
g , and of the energy-momentum tensor of
µν
matter, tM . We understand matter to comprise all the fields
of matter, with the exception of the gravitational field,
µν
tµν = tµν
g + tM .
The interaction between the gravitational field and matter is

taken into account in the density of the energy-momentum
tensor of matter, tµν
M.
Back in 1913 A.Einstein wrote [28]: “... the tensor of the
gravitational field ϑµν is the source of a field together with the
tensor of material systems Θµν . The energy of the gravita-
tional occupying a special position as compared with all other
forms of energy would result in inadmissible consequences”
[28]. We have adopted precisely this idea of A.Einstein as the
basis for constructing the relativistic theory of gravity (RTG).
In constructing general relativity theory (GRT) A.Einstein
was not successful, since instead of the energy-momentum ten-
sor of the gravitational field there arose in GRT the pseu-
dotensor of the gravitational field. All this happened because
29
A.Einstein did not consider the gravitational field a physical
field (such as the Faraday–Maxwell field) in Minkowski space.
Precisely for this reason the equations of GRT do not con-
tain the metric of Minkowski space. From equations (2.2) it
follows that they will also be non-linear for the gravitational
field proper, since the density of the tensor tµν
g is the source of
the gravitational field.
Equations (2.2) and (2.3), which we formally declared the
equations of gravity by analogy with electrodynamics, must
be derived from the principle of least action, since only in
this case we will have an explicit expression for the density
of the energy-momentum tensor of the gravitational field and
of the fields of matter. But, to this end it is necessary to
construct the density of the Lagrangian of matter and of the
gravitational field. Here it is extremely important to realize
this construction on the basis of general principles. Only in
this case one can speak of the theory of gravity. The initial
scalar density of the Lagrangian of matter may be written in
the form
L = Lg (γµν , φ̃µν ) + LM (γµν , φ̃µν , φA ) ,
here Lg is the density of the Lagrangian of the gravitational

field; LM is the density of the Lagrangian of the fields of mat-
ter; φA represents the fields of matter.
The equations for the gravitational field and the fields of
matter have, in accordance with the principle of least action,
the form
δL
=0, (2.4)
δ φ̃µν
δLM
=0. (2.5)
δφA
Equations (2.4) differ from equations (2.2), first of all, in that

the variational derivative of the density of the Lagrangian
there is the derivative with respect to the field φ̃µν , while the
30
variational derivative in equations (2.2) is, in agreement with
definition (2.1), taken from the density of the Lagrangian over
the metric γµν . For equations (2.4) to reduce to equations (2.2)
for any form of matter it is necessary to assume the tensor den-
sity φ̃µν to be always present in the density of the Lagrangian
together with the tensor density γ̃ µν via some common density
g̃ µν in the form
√
g̃ µν = γ̃ µν + φ̃µν , g̃ µν = −gg µν . (2.6)
Thus arises the effective Riemannian space with the

metric g µν (x). Since the gravitational field φ̃µν (x), like
all other physical fields in Minkowski space, can be
described within a sole coordinate map, it is evident
from expression (2.6) that the quantity g̃ µν (x) can also
be fully defined in a sole coordinate map. For descrip-
tion of the effective Riemannian space due to the influence of
the gravitational field, no atlas of maps is required, which is
usually necessary for describing Riemannian space of the gen-
eral form. This means that our effective Riemannian space has
a simple topology. In GRT topology is not simple. Precisely
for this reason, GRT cannot, in principle, be constructed on
the basis of ideas considering gravity a physical gravitational
field in Minkowski space.
If condition (2.6) is taken into account, the density of the
Lagrangian L assumes the form
L = Lg (γµν , g̃ µν ) + LM (γµν , g̃ µν , φA ) .
It must be stressed that condition (2.6) permits substitut-

ing the variational derivative with respect to g̃ µν for the vari-
ational derivative with respect to φ̃µν , and to express the vari-
ational derivative with respect to γµν through the variational
derivative with respect to g̃ µν and the variational derivative
with respect to γµν entering explicitly into the Lagrangian L.
31
Indeed,
δL δL
= µν = 0 , (2.7)
δ φ̃ µν δg̃
%
δL δ L δL ∂g̃ αβ
= + αβ · . (2.8)
δγµν δγµν δg̃ ∂γµν
The derivation of the latter formula is presented in detail in
Appendix (A.17). The asterisk in formula (2.8) indicates the
variational derivative of the density of the Lagrangian with
respect to the metric γµν which is explicitly present in L. In
agreement with (2.1), formula (2.8) can be written in the form
δL ∂g̃ αβ δ%L
tµν = −2 · − 2 .
δg̃ αβ ∂γµν δγµν
Taking equation (2.7) into account in the above expression we
obtain
δ%L
tµν = −2 . (2.9)
δγµν
Comparing equations (2.9) and (2.2) we obtain the condition
δ%L 1 αβ
−2 = [γ Dα Dβ φ̃µν + m2 φ̃µν ] , (2.10)
δγµν 16π
which, in case it is fulfilled, makes it possible to derive the
equations of the gravitational field, (2.2) and (2.3), directly
from the principle of least action. Since the fields of matter
are not present in the right-hand side of (2.10), this means that
the variation in density of the Lagrangian of matter, LM , with
respect to the explicitly present metric γµν must be zero. For
no additional restrictions on the motion of matter determined
by equations (2.5) to arise, it hence follows directly that the
tensor γµν does not explicitly enter into the expression for the
density of the Lagrangian of matter LM . Expression (2.10)
then assumes the form
δ % Lg 1 αβ
−2 = [γ Dα Dβ φ̃µν + m2 φ̃µν ] . (2.11)
δγµν 16π
32
Thus, everything reduces to finding the density of the La-
grangian of the gravitational field proper, Lg , which would
satisfy condition (2.11).
At the same time, from the previous arguments we arrive
at the important conclusion that the density of the Lagrangian
of matter, L, has the form
L = Lg (γµν , g̃ µν ) + LM (g̃ µν , φA ) . (2.12)
Thus, from the requirement that the density of the
energy-momentum tensor of matter be the source of
the gravitational field it follows in a natural way that
the motion of matter should take place in effective Rie-
mannian space. This assertion has the character of a the-
orem. Hence it becomes clear, why Riemannian space arose,
instead of some other. Precisely this circumstance provides
us with the possibility of formulating, in section 3, the gauge
group, and then to construct the density of the Lagrangian
(4.24) satisfying condition (2.11), in accordance with (B.20).
An interesting picture arises consisting in that the motion
of matter in Minkowski space with the metric γµν under the
influence of the gravitational field φµν is identical to the motion
of matter in effective Riemannian space with the metric gµν ,
determined by expression (2.6). We term such interaction of
the gravitational field with matter the g e o m e t i z a t i o n
p r i n c i p l e. The geometrization principle is a consequence of
the initial assumption that a universal characteristic of matter
— the density of the energy-momentum tensor — serves as the
source of the gravitational field. Such a density structure of
the Lagrangian of matter indicates that a unique possibility
is realized for the gravitational field to be attached inside the
Lagrangian density of matter directly to the density of the
tensor γ̃ µν .
The effective Riemannian space is literally of a field
origin, owing to the presence of the gravitational field.
Thus, the reason that the effective space is Riemannian, and
not any other, lies in the hypothesis that a universal conserved
33
quantity — the density of the energy-momentum tensor —
is the source of gravity. We shall explain this fundamental
property of gravitational forces by comparing them with the
electromagnetic forces.
In the case of a homogeneous magnetic field, a charged
particle in Minkowski space is known to undergo, due to the
Lorentz force, motion along a circle in the plane perpendicular
to the magnetic field. However, this motion is far from identi-
cal even for charged particles, if their charge-to-mass ratio dif-
fer. Moreover, there exist neutral particles, and their trajecto-
ries in a magnetic field are just straight lines. Therefore, owing
to the non-universal character of electromagnetic forces their
action cannot be reduced to the geometry of space-time. Gra-
vity is another issue. It is universal, any test bodies move along
identical trajectories given identical initial conditions. In this
case, owing to the hypothesis claiming the energy-momentum
tensor of matter to be the source of the gravitational field,
it turns out to be possible to describe these trajectories by
geodesic lines in the effective Riemannian space-time due to
the presence of the gravitational field in Minkowski space. In
those regions of space, where a whatever small gravitational
field is present, we have metric properties of space which up to
a high precision approach the actually observed properties of
pseudo-Euclidean space. On the other hand, when the gravi-
tational fields are strong, the metric properties of the effective
space become Riemannian. But in this case, also, the pseudo-
Euclidean geometry does not vanish without trace — it is ob-
servable and manifests itself in that the motion of bodies in
effective Riemannian space is not free by inertia, but proceeds
with acceleration with respect to pseudo-Euclidean space in
Galilean coordinates. Precisely for this reason, acceleration in
RTG, unlike GRT, has an absolute sense. Consequently, “Ein-
stein’s lift” cannot serve as an inertial reference frame. This is
manifested in that a charge at rest in “Einstein’s lift” will emit
electromagnetic waves. This physical phenomenon should also
testify in favour of the existence of Minkowski space. As we
34
shall further see, the metric of Minkowski space can be defined
from studies of the distribution of matter and of the motion of
test bodies and light in effective Riemannian space. We shall
raise this issue again in section 7.
The equation of motion of matter does not include the met-
ric tensor γµν of Minkowski space. Minkowski space will only
influence the motion of matter by means of the metric tensor
gµν of Riemannian space, derived, as we shall further see, from
the equations of gravity, which contain the metric tensor γµν of
Minkowski space. Since the effective Riemannian metric arises
on the basis of the physical field given in Minkowski space,
hence it follows that effective Riemannian space has a simple
topology and is presented in a single map. If, for instance,
matter is concentrated in a region of the island-type, then in
Galilean coordinates of an inertial reference system the gra-
vitational field φ̃µν cannot decrease slower than 1/r, but this
circumstance imposes a strong restriction on the asymptotic
behaviour of the metric gµν of effective Riemannian geometry
1
& '
gµν = ηµν + 0 , here ηµν = (1, −1, −1, −1) . (2.13)
r
If, on the other hand, one simply takes as the starting point
the Riemannian metric, without assuming it to have originated
from the action of a physical field, then such restrictions do
not arise, since the asymptotics of the metric gµν even depends
on the choice of three-dimensional space coordinates. Physical
quantity, however, in principle, cannot depend on the choice
of the three-dimensional space coordinates. RTG imposes no
restrictions on the choice of reference system. The reference
system may be arbitrary, if only it realizes a one-to-one cor-
respondence between all the points of the inertial reference
system in Minkowski space and provides for the following in-
equalities, necessary for introducing the concepts of time and
spatial length, to be satisfied:
γ00 > 0, dl2 = sik dxi dxk > 0; i, k = 1, 2, 3,

35
where
γ0i γ0k
sik = −γik + .
γ00
In our theory of gravity the geometrical characteristics of
Riemannian space arise as field quantities in Minkowski space,
and for this reason their transformational properties become
tensor properties, even if this was previously not so, from the
conventional point of view. Thus, for instance, the Christoffel
symbols, given as field quantities in Galilean coordinates of
Minkowski space become tensors of the third rank. In a similar
manner, ordinary derivatives of tensor quantities in Cartesian
coordinates of Minkowski space are also tensors.
The question may arise: why is no division of the metric,
like (2.6), performed in GRT by introduction of the concept of
gravitational field in Minkowski space? The Hilbert-Einstein
equations only contain the quantity gµν , so, consequently, it is
impossible to say unambiguously with the help of which metric
γµν of Minkowski space we should define, in accordance with
(2.6), the gravitational field. But the difficulty consists not
only in the above, but, also, in that the solutions of Hilbert-
Einstein equations are generally found not in one map, but in a
whole atlas of maps. Such solutions for gµν describe Rieman-
nian space with a complex topology, while the Riemannian
spaces, obtained by representation of the gravitational field in
Minkowski space, are described in a sole map and have a simple
topology. It is precisely for these reasons that field representa-
tions are not compatible with GRT, since they are extremely
rigorous. But this means that no field formulation of GRT in
Minkowski space can exist, in principle, no matter how much
someone and who might want this to happen. The apparatus
of Riemannian geometry is inclined towards the possibility of
introducing covariant derivatives in Minkowski space, which
36
we took advantage of in constructing RTG. But to implement
this, it was necessary to introduce the metric of Minkowski
space into the gravitational equations, and it, thus, turned
out to be possible to realize the functional relationship of the
metric of Riemannian space, gµν , with the metric of Minkowski
space, γµν . But this will be dealt with in detail in subsequent
sections.
3. The gauge group of
transformations
Since the density of the Lagrangian of matter has the form
LM (g̃ µν , φA ), (3.1)
it is easy to find the group of transformations, under which

the density of the Lagrangian of matter is only changed by
the divergence. To this end we shall take advantage of the
action #
SM = LM (g̃ µν , φA ) d4 x (3.2)
being invariant under infinitesimal transformations of coordi-
nates,
x! α = xα + ξ α (x), (3.3)
where ξ α is the four-vector of an infinitesimal displacement.
The field functions g̃ µν , φA vary as follows under these trans-
formations of coordinates:
g̃ ! µν (x! ) = g̃ µν (x) + δξ g̃ µν (x) + ξ α(x)Dα g̃ µν (x),
(3.4)
φ!A (x! ) = φA (x) + δξ φA (x) + ξ α (x)Dα φA (x),
where the expressions
δξ g̃ µν (x) = g̃ µα Dα ξ ν (x) + g̃ ναDα ξ µ (x) − Dα (ξ α g̃ µν ),

B;α (3.5)
δξ φA (x) = −ξ α (x)Dα φA (x) + FA;β φB (x)Dα ξ β (x)
are Lie variations.

The operators δξ satisfy the conditions of Lie algebras, i.e.
the commutation relation
[δξ1 , δξ2 ](·) = δξ3 (·) (3.6)
and the Jacobi identity
[δξ1 , [δξ2 , δξ3 ]] + [δξ3 , [δξ1 , δξ2 ]] + [δξ2 , [δξ3 , δξ1 ]] = 0, (3.7)
38
where
ξ3ν = ξ1µ Dµ ξ2ν − ξ2µ Dµ ξ1ν = ξ1µ ∂µ ξ2ν − ξ2µ ∂µ ξ1ν .
For (3.6) to hold valid the following conditions must be satis-
fied:
B; µ C; α B; α C; µ µα; τ C; σ
FA; ν FB; β − FA; β FB; ν = fνβ; σ FA; τ , (3.8)
where the structure constants f are
µα; τ µ α τ α µ τ
fνβ; σ = δβ δσ δν − δν δσ δβ . (3.9)
It is readily verified that they satisfy the Jacobi equality
αν; σ τ ρ; ω νρ; σ τ α; ω ρα; σ τ ν; ω
fβµ; τ fσε; δ + fµε; τ fσβ; δ + fεβ; τ fσµ; δ = 0 (3.10)
and have the property of antisymmetry,
αν; ρ να; ρ
fβµ; σ = −fµβ; σ .
The variation of action under the coordinate transforma-

tion (3.3) equals zero:
# #
4 !
δc S = ! !
L (x ) d x − LM (x) d4 x = 0. (3.11)
Ω! Ω
The first integral in (3.11) can be written in the form

# #
L!M (x! ) d4 x! = J L!M (x! ) d4 x,
Ω! Ω
where $
∂x!α
'
J = det .
∂ xβ
In the first order of ξ α the determinant J equals
J = 1 + ∂α ξ α (x). (3.12)
Taking into account the expansion
∂LM
L!M (x! ) = L!M (x) + ξ α (x) ,
∂xα
39
as well as (3.12), one can represent the expression for the vari-
ation in the form
#
δc SM = [δLM (x) + ∂α (ξ α LM (x))] d4 x = 0.
Ω
Owing to the integration volume Ω being arbitrary, we have

the identity
δ LM (x) = −∂α (ξ α (x)LM (x)), (3.13)
where the Lie variation δLM is

∂LM µν ∂LM
δLM (x) = µν
δg̃ + δ(∂α g̃ µν ) +
∂g̃ ∂(∂α g̃ µν)
∂LM ∂LM
+ δφA + δ(∂α φA ). (3.14)
∂φA ∂(∂α φA )
Hence, for instance, it follows that if the scalar density depends
only on g̃ µν and its derivatives, it will vary under transforma-
tion (3.5) only by the divergence
δL(g̃ µν (x)) = −∂α (ξ α (x) L(g̃ µν (x))), (3.13a)
where the Lie variation δL is

∂L ∂L
δL(g̃ µν (x)) = µν
δg̃ µν + δ(∂α g̃ µν )+
∂g̃ ∂(∂α g̃ µν )
∂L
+ ! δ(∂α ∂β g̃ µν ). (3.14a)
∂(∂α ∂β g̃ µν )
The Lie variations (3.5) were established within the con-

text of the coordinate transformations (3.3). But one may also
adopt another standpoint, in accordance with which transfor-
mations (3.5) can be considered gauge transformations. In this
case an arbitrary infinitesimal four-vector ξ α (x) will already be
a gauge vector, but no longer the displacement vector of the
coordinates. To stress the difference between the gauge group
40
and the group of coordinate transformations, we shall further
use the notation εα (x) for the group parameter and call the
transformation of field functions
g̃ µν (x) → g̃ µν (x) + δg̃ µν (x),

(3.15)
φA (x) → φA (x) + δφA (x)
with the variations
δε g̃ µν (x) = g̃ µα Dα εν (x) + g̃ να Dα εµ (x) − Dα (εα g̃ µν ),

(3.16)
B; α
δε φA (x) = −εα (x)Dα φA (x) + FA; β
β φB (x) Dα ε (x)
gauge transformations.
In full compliance with formulae (3.6) and (3.7), the op-
erators satisfy the same Lie algebra, i.e. the commutation
relation
[δε1 , δε2 ](·) = δε3 (·) (3.17)
and the Jacobi identity
[δε1 , [δε2 , δε3 ]] + [δε3 , [δε1 , δε2 ]] + [δε2 , [δε3 , δε1 ]] = 0. (3.18)
Like in the preceding case, we have
εν3 = εµ1 Dµ εν2 − εµ2 Dµ εν1 = εµ1 ∂µ εν2 − εµ2 ∂µ εν1 .
The gauge group arose from the geometrized structure of

the scalar density of the Lagrangian of matter, LM (g̃ µν , φA ),
which owing to identity (3.13) only changes by the divergence
under gauge transformations (3.16). Thus, the geometrization
principle, which determined the universal character of the in-
teraction of matter and of the gravitational field, has provided
us with the possibility of formulating the non-commutative
infinite-dimensional gauge group (3.16).
The essential difference between the gauge and coordinate
transformations will manifest itself at the decisive point of the
theory in the course of construction of the scalar density of
41
the Lagrangian of the gravitational field proper. The difference
arises owing to the metric tensor γµν not changing under gauge
transformation, and, consequently, owing to (2.6) we have
δε g̃ µν (x) = δε φ̃µν (x).
From (3.16) the transformation for the field follows
δε φ̃µν (x) = g̃ µα Dα εν (x) + g̃ να Dα εµ (x) − Dα (εα g̃ µν ),
but this transformation for the field differs essentially from its
transformation in the case of displacement of the coordinates:
δξ φ̃µν (x) = φ̃µα Dα ξ ν (x) + φ̃να Dα ξ µ (x) − Dα (ξ α φ̃µν ).
Under the gauge transformations (3.16) the equations of mo-

tion for matter do not change, because under any such trans-
formations the density of the Lagrangian of matter is altered
only by the divergence.
4. Density of the Lagrangian and
the equations of motion for the
gravitational field proper
It is known to be impossible, only using the sole tensor gµν , to
construct the scalar density of the Lagrangian of the gravita-
tional field proper with respect to arbitrary coordinate trans-
formations in the form of a quadratic form of derivatives of
order not exceeding the first. Therefore, such a density of the
Lagrangian will certainly contain the metric γµν together with
the metric gµν . But, since the metric γµν is not altered under
the gauge transformation (3.16), there, consequently, must be
imposed strong restrictions on the structure of the density of
the Lagrangian of the gravitational field proper for it to change
only by the divergence under this transformation. It is pre-
cisely here that there arises an essential difference between
gauge and coordinate transformations.
While coordinate transformations impose nearly no restric-
tions on the structure of the scalar density of the Lagrangian of
the gravitational field proper, gauge transformations will per-
mit us to find the density of the Lagrangian. A straightforward
general method for constructing the Lagrangian is presented
in the monograph [10].
Here we shall choose a more simple method for constructing
the Lagrangian. On the basis of (3.13a) we conclude that the
√ √
most simple scalar densities −g and R̃ = −gR, where R is
the scalar curvature of effective Riemannian space, vary as
follows under the gauge transformation (3.16):
√ √ √
−g → −g − Dν (εν −g), (4.1)
R̃ → R̃ − Dν (εν R̃). (4.2)
The scalar density R̃ is expressed via the Christoffel symbols
1 λσ
Γλµν = g (∂µ gσν + ∂ν gσµ − ∂σ gµν ) (4.3)
2
43
as follows:
R̃ = −g̃ µν (Γλµν Γσλσ − Γλµσ Γσνλ ) − ∂ν (g̃ µν Γσµσ − g̃ µσ Γνµσ ). (4.4)
Since the Christoffel symbols are not tensor quantities, no

summand in (4.4) is a scalar density. However, if one intro-
duces the tensor quantities Gλµν
1 λσ
Gλµν = g (Dµ gσν + Dν gσµ − Dσ gµν ), (4.5)
2
then the scalar density can be identically written in the form
R̃ = −g̃ µν (Gλµν Gσλσ −Gλµσ Gσνλ )−Dν (g̃ µν Gσµσ −g̃ µσ Gνµσ ). (4.6)
Note that under arbitrary coordinate transformations each

group of terms in (4.6) individually exhibits the same be-
haviour as scalar density. We see that the apparatus of Rie-
mannian geometry is inclined toward the introduction of co-
variant, instead of ordinary, derivatives in Minkowski space,
but the metric tensor γµν , used for determining the covariant
derivatives, is in no way fixed here.
With account of (4.1) and (4.2), the expression
√
λ1 (R̃ + Dν Qν ) + λ2 −g (4.7)
varies only by the divergence under arbitrary gauge transfor-

mations. Choosing the vector density Qν to be
Qν = g̃ µν Gσµσ − g̃ µσ Gνµσ ,
we exclude from the preceding expression terms containing

derivatives of orders higher, than the first, and obtain the
following density of the Lagrangian:
√
− λ1 g̃ µν (Gλµν Gσλσ − Gλµσ Gσνλ ) + λ2 −g. (4.8)
Thus, we see that the requirement for the density of the

Lagrangian of the gravitational field proper to vary under the
44
gauge transformation (3.16) only by the divergence, unam-
biguously determines the structure of the Lagrangian’s density
(4.8). But, if one restricts oneself only to considering this den-
sity, then the equations of the gravitational field will be gauge
invariant, while the metric of Minkowski space, γµν , will not be
present in the set of equations determined by the density of the
Lagrangian (4.8). Since within such an approach the metric of
Minkowski space disappears, the possibility of representing the
gravitational field as a physical field of the Faraday–Maxwell
type in Minkowski space disappears also.
In the case of the density of the Lagrangian (4.8), intro-
duction of the metric γµν with the aid of equations (2.3) will
not save the situation, since physical quantities — the inter-
val and the curvature tensor of Riemannian space, as well as
the tensor tµν
g of the gravitational field — will depend on the
choice of gauge, which is inadmissible from a physical point of
view. Thus, for example,
δ- Rµν = −Rµσ Dν %σ − Rνσ Dµ %σ − %σ Dσ Rµν ,

δ- Rµναβ = Rσναβ Dµ %σ − Rµσαβ Dν %σ −
−Rµνσβ Dα %σ − Rµνασ Dβ %σ − %σ Dσ Rµναβ .
To retain the concept of a field in Minkowski space and

to exclude the above ambiguity it is necessary to add, in the
density of the Lagrangian of the gravitational field, a term vi-
olating the gauge group. It is precisely here that there arises
an essentially new way, which for a long time evaded being
revealed. At first sight, it may seem that a significant arbi-
trariness should arise here, since the group can be violated in
extremely diverse ways. However, it turns out not to be so,
because our physical requirement, concerning the polarization
properties of the gravitational field which is a field of spins 2
and 0, imposed by equations (2.3), results in the term violat-
ing the group (3.16) being necessarily chosen so as to make
equations (2.3) a consequence of the set of equations of the
gravitational field and of fields of matter, since only in this
45
case we have no over-determined set of differential equations
arising. To this end we introduce into the scalar density of the
Lagrangian of the gravitational field a term of the form
γµν g̃ µν , (4.9)
which, given conditions (2.3), also varies under transforma-

tions (3.16) by the divergence, but only on the class of vectors
satisfying the condition
g µν Dµ Dν εσ (x) = 0. (4.10)
In electrodynamics a nearly analogous situation occurs with

the photon rest mass differing from zero. With account of
(4.8)-(4.9) the general scalar density of the Lagrangian has
the form:
Lg = −λ1 g̃ µν (Gλµν Gσλσ − Gλµσ Gσνλ ) +

√ √
+λ2 −g + λ3 γµν g̃ µν + λ4 −γ. (4.11)
We have introduced the last constant term in (4.11) in order

to use it for reducing to zero the density of the Lagrangian
in absence of the gravitational field. The narrowing of the
class of gauge vectors due to introduction of the term (4.9)
automatically results in equations (2.3) being a consequence
of the equations of the gravitational field. We shall further
verify this directly.
In accordance with the principle of least action, the equa-
tions for the gravitational field proper are of the form
δLg 1
= λ 1 Rµν + λ2 gµν + λ3 γµν = 0, (4.12)
δg̃ µν 2
here $ %
δLg ∂Lg ∂L
= − ∂σ ,
δg̃ µν ∂g̃ µν ∂(∂σ g̃ µν )
where we write the Ricci tensor Rµν in the form
Rµν = Dλ Gλµν − Dµ Gλνλ + Gσµν Gλσλ − Gσµλ Gλνσ . (4.13)

46
Since in absence of the gravitational field equations (4.12) must
be satisfied identically, hence follows
λ2 = − 2 λ3 . (4.14)
Let us now find the density of the energy-momentum for

the gravitational field in Minkowski space
µν δLg √ 1 δLg
t g = −2 2 −γ(γ µα γ νβ − γ µν γ αβ ) αβ +
δ γµν 2 δg̃
+ λ1 J µν µν µν
− 2λ3 g̃ − λ4 γ̃ , (4.15)
where
J µν = Dα Dβ (γ αµ g̃ βν + γ αν g̃ βµ − γ αβ g̃ µν − γ µν g̃ αβ ). (4.16)
(see Appendix (B.19)). If the dynamic equations (4.12) are

taken into account in expression (4.15), then we obtain equa-
tions for the gravitational field proper in the form
λ1 J µν − 2 λ3 g̃ µν − λ4 γ̃ µν = tµν
g . (4.17)
For this equation to be satisfied identically in the absence of

the gravitational field, it is necessary to set
λ4 = −2 λ3 . (4.18)
Since the equality

Dµ tµν
g = 0, (4.19)
always holds valid for the gravitational field proper, from equa-
tion (4.17) it follows that
Dµ g̃ µν = 0. (4.20)
Thus, equations (2.3) determining the polarization states

of the field follow directly from equations (4.17). With account
of equations (4.20), one can write the field equations (4.17) in
the form
λ4 µν 1
γ αβ Dα Dβ φ̃µν − φ̃ = − tµν . (4.21)
λ1 λ1 g
47
In Galilean coordinates this equation has the simple form
λ4 µν 1
φ̃µν − φ̃ = − tµν . (4.22)
λ1 λ1 g
It is natural to consider the numerical factor − λλ41 = m2 to rep-

resent the square graviton mass and to set the value of −1/λ1
equal to 16π, in accordance with the equivalence principle.
Thus, all the unknown constants present in the density of the
Lagrangian have been defined:
1 m2
λ1 = − , λ2 = λ4 = −2 λ3 = . (4.23)
16π 16 π
The constructed scalar density of the Lagrangian of the
gravitational field proper will have the form
1 µν λ σ
Lg = g̃ (Gµν Gλσ − Gλµσ Gσνλ ) −
16 π
m2 1 √ √ '
&
µν
− γµν g̃ − −g − −γ . (4.24)
16 π 2
The corresponding to it dynamic equations for the gravita-
tional field proper can be written down in the form
J µν − m2 φ̃µν = −16π tµν

g , (4.25)
or
m2 µν
Rµν − (g − g µα g νβ γαβ ) = 0. (4.26)
2
These equations impose significant limits on the class of gauge
transformations, retaining only the trivial ones satisfying the
Killing conditions in Minkowski space. Such transformations
are a consequence of Lorentz invariance and are present in any
theory.
The density of the Lagrangian constructed above leads to
equations (4.26) from which it follows that equations (4.20)
are their consequence, and, therefore, outside matter we shall
48
have ten equations for ten unknown field functions. The un-
known field functions φ0α are readily expressed with the aid of
equations (4.20) via the field functions φik , where the indices
i and k run through the values 1, 2, 3.
Thus, the structure of the mass term violating the gauge
group in the density of the Lagrangian of the gravitational
field proper is unambiguously determined by the polarization
properties of the gravitational field. The field approach to
gravity, that declares the energy-momentum tensor of
all matter to be the source of the field, necessarily
requires introduction of the graviton rest mass in the
theory.
5. Equations of motion for the
gravitational field and for matter
The total density of the Lagrangian of matter and of the grav-
itational field is
L = Lg + LM (g̃ µν , φA ), (5.1)
where Lg is determined by expression (4.24).

On the basis of (5.1) we shall obtain, with the aid of the
least action principle, the complete set of equations for matter
and for the gravitational field:
δL
= 0, (5.2)
δg̃ µν
δLM
= 0. (5.3)
δφA
Since in the case of an arbitrary infinitesimal variation of the
coordinates the variation of the action, δc SM , is zero,
#
δc SM = δc LM (g̃ µν , φA ) d4 x = 0,
it is hence possible to obtain an identity (see Appendix (C.16))

in the form
δLM B; ν δLM
& '
gµν ∇λ T λν = −Dν FA; µ φB (x) − Dµ φA (x). (5.4)
δφA δφA
Here T λν = −2 δL M
δgλν
is the density of the tensor of matter in
Riemannian space; ∇λ is the covariant derivative in this space
with the metric gλν . From identity (5.4) it follows that, if
the equations of motion of matter (5.3) are satisfied, then the
following equation occurs:
∇λ T λν = 0. (5.5)
50
When the number of equations (5.3) for matter equals four,
the equivalent equations (5.5) may be used, instead. Since we
shall further only deal with such equations for matter, we shall
always make use of the equations for matter in the form (5.5).
Thus, the complete set of equations for matter and for the
gravitational field will have the form
δL
= 0 , (5.6)
δg̃ µν
∇λ T λν = 0. (5.7)
Matter will be described by velocity 0v, the density of matter

ρ, and pressure p. The gravitational field will be determined
by ten components of the tensor φµν .
Thus, we have 15 unknowns. For determining them it is
necessary to add to the 14 equations (5.6),(5.7) the equation of
state for matter. If the relations (see Appendices B∗ .18,B∗ .19)
δLg 1 m2
= − R µν + (gµν − γµν ), (5.8)
δg̃ µν 16 π 32 π
δLM 1 1
& '
= √ Tµν − gµν T , (5.9)
δg̃ µν 2 −g 2
are taken into account, then the set of equations (5.6), (5.7)
may be represented as
m2 µν
(
1
& '
Rµν − g µν R + g + (g µα g νβ −
2 2
)
1 µν αβ 8 π µν
− g g ) γαβ = √ T , (5.10)
2 −g
∇λ T λν = 0. (5.11)
Owing to the Bianchi identity
1 µν
∇µ (Rµν − g R) = 0
2
51
from equations (5.10) we have
√ 1
m2 −g(g µα g νβ − g µν g αβ )∇µ γαβ = 16 π ∇µ T µν . (5.12)
2
Taking into account expression
∇µ γαβ = −Gσµα γσβ − Gσµβ γσα , (5.13)
where Gσµα is defined by formula (4.5), we find
1 µν αβ
(g µα g νβ − g g ) ∇µ γαβ =
2
= γµλ g µν (Dσ g σλ + Gβαβ g αλ ), (5.14)
but since (see formulae (B∗.20))
√
−g(Dσ g σλ + Gβαβ g αλ ) = Dσ g̃ λσ , (5.15)
expression (5.14) assumes the form
√ 1 µν αβ
g g ) ∇µ γαβ = γµλ g µν Dσ g̃ λσ . (5.16)
−g(g µα g νβ −
2
With the aid of (5.16) expression (5.12) can be represented in
the form
m2 γµλ g µν Dσ g̃ λσ = 16π ∇µ T µν .
This expression can be rewritten in the form
m2 Dσ g̃ λσ = 16 π γ λν ∇µ Tνµ . (5.17)
With the aid of this relation, equation (5.11) can be replaced
by the equation
Dσ g̃ νσ = 0. (5.18)
Therefore, the set of equations (5.10), (5.11) is reduced to the
set of gravitational equations in the form
1 m2 * µν
& '
µν
R − g µν R + g + (g µα g νβ −
2 2
1 8 π µν
+
− g µν g αβ )γαβ = √ T , (5.19)
2 −g
Dµ g̃ µν = 0. (5.20)
52
These equations are universally covariant with respect to
arbitrary transformations of coordinates and form-invariant
only with respect to such transformations of coordinates that
leave the Minkowski metric γµν (x) form-invariant. Hence, for
instance, it follows that in any inertial (Galilean) reference sys-
tem phenomena are described by identical equations. Equa-
tions involving the graviton mass had arisen previously; how-
ever, owing to misunderstanding of the fundamental fact that
special relativity theory is also valid in non-inertial reference
systems, they were not considered seriously, since they were
not universally covariant. Usually, following A.Einstein, the
metric ηαβ = (1, −1, −1, −1) was considered to be a tensor
only with respect to the Lorentz transformations. But, actu-
ally, the metric of Minkowski space, γµν (x), is a tensor with
respect to arbitrary transformations of coordinates. The set of
equations (5.19) and (5.20) is hyperbolic. In the case of static
problems, it is elliptic. By adding the equation of state to the
set of equations (5.19) and (5.20) we obtain a complete set
of equations for determining the unknown physical quantities
gµν , 0v, ρ, p for one or another formulations of the problem.
A concrete inertial Galilean reference system is singled out
by formulation of the actual physical problem (by the initial
and boundary conditions). The descriptions of a given formu-
lated physical problem in different inertial (Galilean) reference
systems are different, naturally, but this does not contradict
the relativity principle. If one introduces the tensor
m2 µν
N µν = Rµν − [g − g µα g νβ γαβ ], N = N µν gµν ,
2
then the set of equations (5.19) and (5.20) can be written in
the form
1 8 π µν
N µν − g µν N = √ T , (5.19a)
2 −g
Dµ g̃ µν = 0. (5.20a)
53
It may also be represented in the form
8π 1
N µν = √ (T µν − g µν T ), (5.21)
−g 2
Dµ g̃ µν = 0, (5.22)
or
8π 1
Nµν = √ (Tµν − gµν T ), (5.21a)
−g 2
Dµ g̃ µν = 0. (5.22a)
It must be especially stressed that both sets of equations
(5.21) and (5.22) contain the metric tensor of Minkowski space.
Transformations of coordinates, which leave the metric of
Minkowski space form-invariant, relate physically equivalent
reference systems. The most simple of these are inertial refer-
ence systems. For this reason, possible gauge transformations
satisfying the Killing conditions
Dµ εν + Dν εµ = 0,
do not remove us from the class of physically equivalent refer-

ence systems.
Let us deal with this issue in more detail. To this end we
write the equation of RTG, (5.21) and (5.22), in the expanded
form:
m2 µν
R µν (x) − [g (x) − g µα g νβ γαβ (x)] =
2
1 µν
, +
µν
= 8π T (x) − g T (x) , (5.23)
2
Dµ g̃ µν = 0. (5.24)
Consider that, given appropriate conditions of the problem,
these equations have the solution g µν (x) in Galilean coordi-
nates x in an inertial reference system, when the distribution
54
of matter is T µν (x). In another inertial reference system, in
Galilean coordinates x! satisfying the condition
x!ν = xν + %ν (x),
(5.25)
D µ %ν + D ν %µ = 0.
We obtain with the aid of tensor transformations the following:
m2 !µν !
R !µν (x! ) −[g (x ) − g !µα g !νβ γαβ (x! )] =
2
1 !µν ! !
, +
!µν !
= 8π T (x ) − g T (x ) . (5.26)
2
Since equations (5.23) are form-invariant with respect to
the Lorentz transformations, we can return to the initial vari-
ables x:
m2 !µν
R !µν (x) − [g (x) − g !µα g !νβ γαβ (x)] =
2
1 !µν !
, +
!µν
= 8π T (x) − g T (x) (5.27)
2
Hence it is clear that the solution g !µν (x) does not correspond
to the distribution of matter T µν (x), but to another distribu-
tion T !µν (x). The quantity g !µν (x) in equations (5.27) is
g !µν (x) = g µν (x) + δ- g µν , (5.28)
where
δ- g µν = g µλ Dλ %ν + g νλ Dλ %µ − %λ Dλ g µν . (5.29)
In the case of transformations (5.25) we have
R!µν (x) = Rµν (x) + δ- Rµν ,

T !µν (x) = T µν (x) + δ- T µν , (5.30)
T ! (x) = T (x) + δ- T.
55
Here
δ- Rµν = Rµλ Dλ %ν + Rνλ Dλ %µ − %λ Dλ Rµν ,

δ- T µν = T µλ Dλ %ν + T νλ Dλ %µ − %λ Dλ T µν , (5.31)
δ- T = −%λ Dλ T = −%λ ∂λ T.
We obtained expression (5.27) with the aid of the coordi-

nate transformations (5.25), but identical equations are also
obtained in the case of the gauge transformation (5.29) with
the vectors %λ satisfying condition (5.25). Thus, gauge trans-
formations result in the metric field g !µν (x) in the case of
matter exhibiting the distribution T !µν (x). Although we con-
sidered the transition from one inertial reference system in
Galilean coordinates to another, our formulae (5.25), (5.31)
are of a general nature, they are valid, also, for a noniner-
tial reference system in Minkowski space. The same situation
occurs in electrodynamics.
In GRT the situation is completely different, since owing
to the Hilbert–Einstein equations being form-invariant with
respect to arbitrary transformations of coordinates, there ex-
ist, for one and the same distribution of matter, T µν (x), any
!
amount of metrics gµν (x), gµν (x)... satisfying the equations. It
is precisely for this reason that in GRT there arises an ambi-
guity in the description of gravitational phenomena.
If one imagines that it is possible to perform experimental
measurements of the characteristics of Riemannian space and
of the motion of matter with whatever high precision, then
it would be possible, on the basis of equations (5.21a) and
(5.22a), to determine the metric of Minkowski space and to
find Galilean (inertial) reference systems. Thus, Minkowski
space is observable, in principle.
The existence of Minkowski space is reflected in the con-
servation laws, and, therefore, testing their validity in physical
phenomena serves at the same time for testing the structure
of space-time.
It must be especially noted that both sets of equations
56
(5.19) and (5.20) contain the metric tensor of Minkowski space.
The presence of the cosmological term in the equations of GRT
is known not to be obligatory, and this issue is still being dis-
cussed. The presence of the cosmological term in the equations
of gravity is obligatory in RTG. However, the cosmological
term in equations (5.19) arises in combination with the term
related to the metric γµν of Minkowski space, and with the
same constant factor equal to a half of the square graviton
mass. The presence in equations (5.19) of the term connected
with the metric γµν significantly alters the character of the
collapse and development of the Universe. In accordance with
equations (5.19), in absence of matter and of the gravitational
field, the metric of the space becomes the Minkowski metric,
and it coincides exactly with the one previously chosen in for-
mulating the physical problem. If the metric of Minkowski
space were absent in the equations of the gravitational field,
then it would be absolutely unclear, in which reference system
of Minkowski space we would happen to be in the absence of
matter and of the gravitational field.
The graviton mass is essential for this theory, since only its
introduction permits construction of the theory of gravity in
Minkowski space. The graviton mass violates the gauge group
or, in other words, it removes the degeneracy. Therefore, one
cannot exclude the possibility of the graviton mass tending
toward zero in the final results, when gravitational effects are
studied. However, the theory with a graviton mass and the
theory involving violation of the gauge group [8] (with the
graviton mass subsequently tending toward zero) are essen-
tially different theories. Thus, for example, while in the first
theory the Universe is homogeneous and isotropic, no such
Universe can exist in the second one.
Let us now touch upon the equivalence principle. Any
physical theory must comply with the equivalence principle.
Gravitational interactions alter the equations of motion of
matter. The requirement imposed by the equivalence prin-
ciple reduces to the requirement that, when the gravitational
57
interaction is switched off, i.e. when the Riemann curvature
tensor turns to zero, these equations of motion become or-
dinary equations of motion of SRT in the chosen reference
system.
In formulating the physical problem in RTG we choose a
certain reference system with the metric tensor of Minkowski
space, γµν (x). In RTG, the equation of motion of matter in
effective Riemannian space with the metric tensor gµν (x), de-
termined by the equations of the gravitational field (5.19) and
(5.20), has the form
∇µ T µν (x) = 0. (σ)
For example, we shall take dustlike matter with the energy-

momentum tensor T µν equal to
µν µ ν dxν
ν
T (x) = ρU U , U = ,
ds
ds is the interval in Riemannian space.
On the basis of equations (σ), using the expression for T µν ,
we find the equation for the geodesic line in Riemannian space,
dU ν
+ Γναβ (x)U α U β = 0.
ds
When the gravitational interaction is switched off, i.e. when
the Riemann curvature tensor turns to zero, from the equa-
tions of the gravitational field (5.19) and (5.20) it follows that
the Riemannian metric gµν (x) transforms into the previously
chosen metric of Minkowski space, γµν (x). In this case the
equation of motion of matter (σ) assumes the form
Dµ tµν (x) = 0. (λ)
Here the energy-momentum tensor tµν (x) is

dxν
tµν (x) = ρuµ uν , uν = ,
dσ
58
dσ is the interval in Minkowski space.
On the basis of (λ), using the expression for tµν , we find
the equations for the geodesic line in Minkowski space,
duν ν
+ γαβ uα uβ = 0,
dσ
i.e., we have arrived at the ordinary equations for the free
motion of particles in SRT in the previously chosen reference
system with the metric tensor γµν (x). Thus, the equation of
motion of matter in a gravitational field in the chosen refer-
ence system automatically transforms, when the gravitational
interaction is switched off, i.e., when the Riemann curvature
tensor turns to zero, into the equation of motion of matter in
Minkowski space in the same reference system with the metric
tensor γµν (x), i.e., the equivalence principle is obeyed. This
assertion in RTG is of a general character, since when the Rie-
mann tensor turns to zero the density of the Lagrangian of
matter in the gravitational field, LM (g̃ µν , ΦA ), transforms into
the ordinary SRT density of the Lagrangian, LM (γ̃ µν , ΦA ), in
the chosen reference system.
In GRT the equation of motion of matter also has the form
(σ). But, since the Hilbert–Einstein equations do not contain
the metric tensor of Minkowski space, then when the gravita-
tional interaction is switched off, i.e., the Riemann curvature
tensor turns to zero, it is impossible to say in which reference
system (inertial or accelerated) of Minkowski space we happen
to be, and therefore it is impossible to determine which equa-
tion of motion of matter in Minkowski space we will obtain
when the gravitational interaction is switched off. Precisely
for this reason, the equivalence principle cannot be complied
with in GRT within the framework of this theory. Usually
such correspondence in GRT is achieved precisely within the
field approach, when a weak gravitational field is considered to
be a physical field in Minkowski space in Galilean coordinates.
Thus, GRT is made to include what it essentially does not con-
tain, since, as A. Einstein wrote, “gravitational fields can be
59
given without introducing tensions and energy density” 20 , so
no existence of any physical field in GRT can even be spoken
of.
In conclusion, we note that RTG revives all those concepts
(inertial reference system, the law of inertia, acceleration rela-
tive to space, the conservation laws of energy-momentum and
of angular momentum), which occurred in classical Newtonian
mechanics and in special relativity theory, and which had to
be renounced by A. Einstein in constructing GRT.
In 1955 A.Einstein wrote: ¡¡A significant achievement of
general relativity theory consists in that it rids physics of the
necessity of introducing the “inertial system” (or “inertial sys-
tems”)¿¿ 21 . From our point of view, the fields of inertia and of
gravity cannot be identified with each other even locally, since
they are of totally different natures. While the former can be
removed by a choice of the reference system, no choice what-
ever of the reference system can remove the fields of gravity.
Regrettably, this circumstance is not understood by many per-
sons, since they do not apprehend that “in Einstein’s theory”,
as especially stressed by J.L.Synge, “the existence or absence
of a gravitational field depends on whether the Riemann cur-
vature tensor differs from or equals zero” 22.
20
vol.1, art.47, p.627.
21
vol.2, art.146, p.854.
22
J.L.Synge. Relativity: the general theory. M.:Foreign literature
6. The causality principle in RTG
RTG was constructed within the framework of SRT, like the
theories of other physical fields. According to SRT, any motion
of a pointlike test body (including the graviton) always takes
place within the causality light cone of Minkowski space. Con-
sequently, non-inertial reference systems, realized by test bod-
ies, must also be inside the causality cone of pseudo-Euclidean
space-time. This fact determines the entire class of possible
non-inertial reference systems. Local equality between the
three-dimensional force of inertia and gravity in the case of
action on a material pointlike body will occur, if the light cone
of the effective Riemannian space does not go beyond the lim-
its of the causality light cone of Minkowski space. Only in
this case can the three-dimensional force of the gravitational
field acting on the test body be locally compensated by transi-
tion to the admissible non-inertial reference system, connected
with this body.
If the light cone of the effective Riemannian space were to
reach beyond the causality light cone of Minkowski space, this
would mean that for such a “gravitational field” no admissible
non-inertial reference system exists, within which this “force
field” could be compensated in the case of action on a material
pointlike body. In other words, local compensation of the
3-force of gravity by the force of inertia is possible only when
the gravitational field, acting as a physical field on particles,
does not lead their world lines outside the causality cone of
pseudo-Euclidean space-time. This condition should be con-
sidered the causality principle permitting selection of solutions
of the set of equations (5.19) and (5.20) having physical sense
and corresponding to the gravitational fields.
The causality principle is not satisfied automatically. There
is nothing unusual in this fact, since both in electrodynamics,
and in other physical theories, as well, the causality condi-
tion for matter in the form dσ 2 = γµν dxµ dxν ≥ 0 is always
61
added (but not always noted) to the main equations, which
actually provides for it being impossible for any form of mat-
ter to undergo motion with velocities exceeding the speed of
light. In our case it is necessary to take into account that
the gravitational interaction enters into the coefficients of the
second-order derivatives in the field equations, i.e. there arises
an effective geometry of space-time. This feature is only pe-
culiar to the gravitational field. The interaction of all other
known physical fields usually does not involve the second-order
derivatives of the field equations, and therefore does not alter
the initial pseudo-Euclidean geometry of space-time.
We shall now present an analytical formulation of the ca-
usality principle in RTG. Since in RTG the motion of matter
under the action of the gravitational field in pseudo-Euclidean
space-time is equivalent to the motion of matter in the corre-
sponding effective Riemannian space-time, we must for events
(world lines of particles and of light) related by causality, on
the one hand, have the condition
d s2 = gµν dxµ dxν ≥ 0, (6.1)
and, on the other hand, the following inequality must hold
valid for such events:
dσ 2 = γµν dxµ dxν ≥ 0. (6.2)
The following condition is valid for the chosen reference system
realized by physical bodies:
γ00 > 0. (6.3)
We single out in expression (6.2) the time- and spacelike parts:
γ0 i dxi
'2
√
&
2
dσ = γ00 dt + √ − si k dxi dxk , (6.4)
γ00
here the Latin indices i, k run through the values 1, 2, 3;
γ0 i γ0 k
si k = − γ i k + , (6.5)
γ00
62
si k is the metric tensor of three-dimensional space in four-
dimensional pseudo-Euclidean space-time. The square spatial
distance is determined by the expression
d l2 = si k dxi dxk . (6.6)

i
Now we represent the velocity v i = dx
dt
as v i = vei , where
v is the absolute value of the velocity and ei is an arbitrary
unit vector in three-dimensional space,
si k ei ek = 1. (6.7)
In absence of the gravitational field the velocity of light in the

chosen reference system is readily determined from expression
(6.4) by setting it equal to zero:
γ0i dxi
'2
√
&
γ00 dt + √ = si k dxi dxk .
γ00
Hence, we find
√ γ0i ei
& '
v= γ00 / 1 − √ . (6.8)
γ00
Thus, an arbitrary four-dimensional isotropic vector in Min-
kowski space, uν , is
uν = (1, v ei ). (6.9)
For both conditions (6.1), (6.2) to be satisfied simultane-

ously, it is necessary and sufficient that for any isotropic vector
γµν uµ uν = 0 (6.10)
the causality condition
gµν uµ uν ≤ 0, (6.11)
hold valid, which precisely indicates that the light cone of the
effective Riemannian space does not go beyond the causality
63
light cone of pseudo-Euclidean space-time. The causality con-
dition may be written in the following form:
gµν v µ v ν = 0, (6.10a)
γµν v µ v ν ≥ 0. (6.11a)
In GRT, physical meaning is also attributed to such solu-

tions of the Hilbert-Einstein equations, which satisfy the in-
equality
g < 0,
as well as the requirement known as the energodominance con-
dition, which is formulated as follows. For any timelike vector
Kν the inequality
T µν Kµ Kν ≥ 0,
must be valid, and the quantity T µν Kν must form, for the
given vector Kν , a non-spacelike vector.
In our theory, such solutions of equations (5.21a) and (5.22a)
have physical meaning, which, besides these requirements, must
also satisfy the causality condition (6.10a) and (6.11a). The
latter can be written, on the basis of equation (5.21a), in the
following form:
8π 1
Rµν K µ K ν ≤ √ (Tµν − gµν T ) K µ K ν +
−g 2
m2
+ gµν K µ K ν . (6.12)
2
If the density of the energy-momentum tensor is taken in the
form:
√
Tµν = −g[(ρ + p)Uµ Uν − pgµν ],
then on the basis of (5.21a) it is possible to establish between

the interval of Minkowski space, dσ, and the interval of the
64
effective Riemannian space, ds, the following relationship:
m2 2 2 m2
dσ = ds [4π(ρ + 3p) + − Rµν U µ U ν ],
2 2
dxµ
here U µ = .
ds
Owing to the causality principle the inequality
m2
Rµν U µ U ν < 4π(ρ + 3p) + ,
2
which is a special case of inequality (6.12), or
√
−gRµν v µ v ν ≤ 8 πTµν v µ v ν (6.12a)
must hold valid.

Let us now consider the motion of a test body under the in-
fluence of gravity in GRT and RTG. In 1918 A. Einstein gave
the following formulation of the equivalence principle: “In-
ertia and gravity are identical; hence and from the results of
special relativity theory it inevitably follows that the symmetric
'fundamental tensor( gµν determines the metric properties
of space, of the motion of bodies due to inertia in it, and,
also, the influence of gravity” 23 . Identifying in GRT the grav-
itational field and the metric tensor gµν of Riemannian space
permits, by an appropriate choice of the reference system, to
equate to zero all the components of the Christoffel symbol at
all points of an arbitrary non-selfintersecting line. Precisely
for this reason, motion along a geodesic line in GRT is con-
sidered free. But, in this case, the choice of reference system
cannot remove the gravitational field in GRT, also, because
the motion of two close material pointlike bodies will not be
free due to the existence of the curvature tensor, which can
never be equated to zero by a choice of reference system owing
to its tensor properties.
23
vol.1, art.45, p.613.
65
The gravitational field in RTG is a physical field in the
spirit of Faraday–Maxwell, so the gravitational force is de-
scribed by a four-vector, and, consequently, the forces of iner-
tia can be made to compensate the three-dimensional part of
the force of gravity by a choice of reference system only if con-
ditions (6.10) and (6.11) are satisfied. Now, the motion of a
material pointlike body in the gravitational field can never be
free. This is especially evident, if the equation of the geodesic
line is written in the form [41]
DU ν
= −Gραβ U α U β (δρν − U ν Uρ ).
dσ
Here
dxν
dσ 2 = γµν dxµ dxν , U ν = .
dσ
Free motion in Minkowski space is described by the equa-
tion:
DU ν dU ν ν
= + γµλ U µ U λ = 0,
dσ dσ
ν
γµλ are the Christoffel symbols of Minkowski space. We see
that motion along a geodesic line of Riemannian space is the
motion of a test body under the action of the force F ν :
F ν = −Gραβ U α U β (δρν − U ν Uρ ),
and this force is a four-vector. If the test body were charged,

it would emit electromagnetic waves, since it moves with ac-
celeration.
In SRT there exists an essential difference between the
forces of inertia and physical forces. The forces of inertia can
always be made equal to zero by a simple choice of reference
system, while essentially no choice of reference system can turn
physical forces into zero, since they are of a vector nature in
Minkowski space. Since in RTG all forces, including gravi-
tational forces, are of a vector nature, this means that they
cannot be equated to zero by a choice of reference system. A
choice of reference system can only make the force of inertia
66
compensate a three-dimensional force, acting on a material
pointlike body, the force being of any nature, including gravi-
tational. In GRT, as noted by J.L.Synge [23], “... the concept
of the force of gravity does not exist, since gravitational prop-
erties are organically present in the structure of space-time
and are manifested in the curvature of space-time, i.e. in that
the Riemann tensor Rijkm differs from zero.” Precisely in this
connection, J.L.Synge wrote: “According to the famous leg-
end, Newton was inspired to create his theory of gravity, when
he once observed an apple falling from the branch of a tree,
and those who study Newtonian physics even now are ready
to claim that the acceleration (980 cm/s2 ) of a falling apple
is due to the gravitational field. In accordance with relativity
theory (GRT is intended — A.L.), this point of view is com-
pletely erroneous. We shall undertake a thorough investigation
of this problem and verify that the gravitational field (i.e., the
Riemann tensor) actually plays an extremely insignificant role
in the phenomenon of a falling apple, while the acceleration
980 cm/s2 is really due to the curvature of the world line of
the tree’s branch.”
According to RTG, the gravitational field is a physical field,
and therefore, unlike the case of GRT, it fully retains the con-
cept of the force of gravity. Precisely owing to the force of
gravity does the free fall of bodies occur, i.e., everything pro-
ceeds like in Newtonian physics. Moreover, all gravitational
effects in the Solar system (the displacement of the perihelion
of Mercury, the deflection of light by the Sun, the time de-
lay of a radiosignal, the precession of a gyroscope) are caused
precisely by the action of the force of gravity, but not by the
curvature tensor of space-time, which in the Solar system is
quite small.
The local identity between inertia and gravity was seen by
Einstein as the main reason for the inertial and gravitational
masses to be equal to each other. However, in our opinion, as
it can be seen from equations (2.2), the reason for this equality
lies in that the source of the gravitational field is the conversed
67
total density of the tensor of matter and of the gravitational
field. Precisely for this reason, the inertial and gravitational
masses being equal to each other does not require the forces
of gravity and of inertia to be locally identical.
7. Mach’s principle
In formulating the laws of mechanics Newton introduced the
notion of absolute space, which always remains the same and
is motionless. He defined the acceleration of a body precisely
with respect to this space. This acceleration had an absolute
character. The introduction of such an abstraction as abso-
lute space turned out to be extremely fruitful. Hence, for
instance, arose the concepts of inertial reference systems in
the entire space, the relativity principle for mechanical pro-
cesses, and the idea came into being of states of motion, that
are physically singled out. In this connection Einstein wrote
the following in 1923: “Reference systems that are in such
states of motion are characterized by the laws of Nature for-
mulated in these coordinates assuming the most simple form.”
And further: ¡¡...according to classical mechanics there exists
“relativity of velocity”, but not “relativity of acceleration”¿¿ 24.
Thus in theory was the notion established of inertial refer-
ence systems, in which material pointlike bodies, not subject to
the action of forces, do not experience acceleration and remain
at rest or in their state of uniform motion along a straight line.
However, Newton’s absolute space or inertial reference system
were actually introduced a priori, without any relation to the
character of the distribution of matter in the Universe.
Mach displayed much courage in seriously criticizing the
main points of Newton’s mechanics. He later wrote that he
succeeded in publishing his ideas with difficulty. Although
Mach did not construct any physical theory free of the dis-
advantages he himself pointed out, he greatly influenced the
development of physical theory. He drew the attention of sci-
entists to the analysis of the main physical concepts.
We shall quote some statements made by Mach [18], which
in the literature have been termed the “Mach principle”. “No
24
vol.2, art.70, p.122.
69
one can say anything about absolute space and absolute mo-
tion, this is only something that can be imagined and is not
observable in experiments”. And further: “Instead of referring
a moving body to space (to some reference system), we shall
directly consider its relation to b o d i e s of the world, only
by which it is possible to d e f i n e a reference system. ...even
in the most simple case, when we apparently consider the in-
teraction between only t w o masses, it is i m p o s s i b l e
to become distracted from the rest of the world. ... If a body
revolves with respect to the sky of motionless stars, then there
arise centrifugal forces, while if it revolves round a n o t h e r
body, instead of the sky of motionless stares, no centrifugal
forces will arise. I have nothing against calling the first rev-
olution a b s o l u t e, if only one does not forget that this
signifies nothing but revolution r e l a t i v e to the sky of
motionless stars.”
Therefore Mach wrote: “...there is no necessity for relating
the Law of inertia to some special absolute space. The most
natural approach of a true naturalist is the following: first to
consider the law of inertia as quite an approximate law, then
to establish its relationship in space to the motionless sky of
stars, ...and then one should expect corrections or some devel-
opment of our knowledge on the basis of further experiments.
Not long ago Lange published a critical article, in which he
exposes how it would be possible, in accordance with his prin-
ciples, to introduce a n e w reference system, if the ordinary
rough reference to the motionless starry sky were to become no
longer suitable owing to more precise astronomical observa-
tions. There exists no difference between the opinion of Lange
and my own relative to the t h e o r e t i c a l formal value
of Lange’s conclusions, namely, that at present the motionless
starry sky is the only p r a c t i c a l l y suitable reference
system, and, also, relative to the method of defining a new
reference system by gradually introducing corrections.” [18].
Further, Mach quotes S. Neumann:“Since all motions must
be referred to the reference system alpha (the reference sys-
70
tem of inertia), it evidently represents an indirect relationship
between all the processes taking place in the Universe, and,
consequently, it contains, so to say, a universal law which is
just as mysterious as it is complex”. In this connection Mach
notes: “I think anyone will agree with this” [18].
From Mach’s statements it is obvious that, since the issue
concerns the law of inertia, in accordance with which, following
Newton, “...each individual body, being left to itself, retains its
state of rest or uniform motion along a straight line...”, there
naturally arises the question of inertial reference systems and
of their relations to the distribution of matter. Mach and his
contemporaries quite clearly understood that such a relation
should exist in Nature. Precisely this meaning will further be
attributed to the concept of “Mach’s principle”.
Mach wrote: “Although I think that at the beginning as-
tronomical observations will necessitate only very insignificant
corrections, I anyhow do think it possible that the law of iner-
tia in the simple form given it by Newton plays for us, human
beings, only a limited and transient role.” [18]. As we shall
further see, Mach did not turn out to be right, here. Mach did
not give a mathematical formulation of his idea, and there-
fore very often diverse authors attribute to Mach’s principle
diverse meanings. We shall try, here, to retain the meaning,
attributed to it by Mach himself.
Poincaré, and later Einstein, generalized the relativity prin-
ciple to all physical phenomena. Poincaré’s formulation [40]
goes as follows: “...the relativity principle, according to which
the laws governing physical phenomena should be identical for
an observer at rest and for an observer undergoing uniform
motion along a straight line, so we have and can have no
method for determining whether we are undergoing similar
motion or not.” Application of this principle to electromag-
netic phenomena led Poincaré, and then Minkowski, to the
discovery of the pseudo-Euclidean geometry of space-time and
thus even more reinforced the hypothesis of inertial reference
systems existing throughout the entire space. Such reference
71
systems are physically singled out, and therefore acceleration
relative to them has an absolute sense.
In general relativity theory (GRT) no inertial reference sys-
tems exist in all space. Einstein wrote about this in 1929:
“The starting point of theory is the assertion that there ex-
ists no singled out state of motion, i.e. not only velocity, but
acceleration has no absolute sense” 25 .
Mach’s principle, in his own formulation, turned out not
to have any use. It must, however, be noted that the ideas of
inertial reference systems throughout the space have quite a
weighty basis, since, for instance, in passing from a reference
system bound to the Earth to a reference system bound to the
Sun and, then, further to the Metagalaxy we approach, with
an increasing precision, the inertial reference system. There-
fore, there are no reasons for renouncing such an important
concept as the concept of an inertial reference system. On
the other hand, the existence of the fundamental conserva-
tion laws of energy-momentum and of angular momentum also
leads with necessity to the existence of inertial reference sys-
tems in the entire space. The pseudo-Euclidean geometry of
space reflects the general dynamic properties of matter and
at the same time introduces inertial reference systems. Al-
though the pseudo-Euclidean geometry of space-time resulted
from studies of matter, and therefore cannot be separated from
it, nevertheless, it is possible to speak of Minkowski space in
the absence of matter. However, like earlier in Newtonian me-
chanics, in special relativity theory no answer exists to the
question of how inertial reference systems are related to the
distribution of matter in the Universe.
The discovery of the pseudo-Euclidean geometry of space
and time permitted considering not only inertial, but accelera-
ted reference systems, also, from a unique standpoint. A large
difference was revealed between the forces of inertia and forces
25
vol.2, art.92, p.264.
72
caused by physical fields. It consists in that the forces of inertia
can always be equated to zero by choosing an appropriate ref-
erence system, while forces caused by physical fields cannot,
in principle, be made equal to zero by a choice of reference
system, since they are of a vector nature in four-dimensional
space-time. Since the gravitational field in RTG is a physical
field in the spirit of Faraday-Maxwell, forces caused by such a
field cannot be equated to zero by a choice of reference system.
Owing to the gravitational field having a rest mass, the
main equations of RTG, (5.19) and (5.20), contain, together
with the Riemannian metric, the metric tensor of Minkowski
space, also, but this means that, in principle, the metric of this
space can be expressed via the geometric characteristics of the
effective Riemannian space and, also, via quantities charac-
terizing the distribution of matter in the Universe. This is
readily done by passing in equations (5.19) from contravariant
to covariant quantities. In this way we obtain
m2 8π 1 m2
γµν (x) = √ (Tµν − gµν T ) − Rµν + gµν . (7.1)
2 −g 2 2
Hence, we see that in the right-hand side of the equations
there occur only geometric characteristics of the effective Rie-
mannian space and quantities determining the distribution of
matter in this space.
Experimental investigation of the motion of particles, and
of light, in Riemannian space, in principle, allows to find the
metric tensor of Minkowski space and, consequently, to con-
struct an inertial reference system, also. Thus, RTG con-
structed within the framework of special relativity theory per-
mits to establish the relation between an inertial reference sys-
tem and the distribution matter. For this reason, motion rel-
ative to space is motion relative to matter in the Universe.
The existence of an inertial reference system, determined by
the the distribution of matter, makes acceleration absolute.
We see that the special relativity principle is of general signif-
icance, independent of the form of matter.
73
The requirements of this principle in the case of the grav-
itational field are expressed by the condition that equations
(5.19) and (5.20) be form-invariant relative to the Lorentz
group. Lorentz form-invariance of physical equations remains
a most important physical principle in constructing a theory,
since precisely this principle provides the possibility of intro-
ducing universal characteristics for all forms of matter.
A. Einstein wrote in his work of 1950: “...should one not fi-
nally try to retain the concept of an inertial system, renouncing
all attempts at explaining the fundamental feature of gravita-
tional phenomena, which manifests itself in Newton’s system
as the equivalence of inert and gravitating masses?” 26. The
concept of an inertial system is retained in RTG, and at the
same time it is shown that the equivalence of inert and grav-
itating masses is a direct consequence of the hypothesis that
the conserved density of the energy-momentum tensor of mat-
ter is the source of the gravitational field. Thus, the equality
between inert and gravitating masses in no way contradicts the
existence of an inertial reference system. Moreover, these con-
ditions organically complement each other and underlie RTG.
Contrary to our conclusion, A.Einstein gave the following
answer to his own question: “Who believes in the compre-
hensibility of Nature should answer — no.” The existence of
inertial reference systems permits resolving Mach’s paradox,
since only in this case can one speak of acceleration relative
to space. V.A.Fock wrote in this connection: ¡¡As to Mach’s
paradox, it is known to be based on the consideration of a rotat-
ing liquid, having the shape of an ellipsoid, and of a spherical
body that does not rotate. The paradox arises, here, only if the
concept “rotation relative to space” is considered to be sense-
less; then, indeed, both bodies (the rotating one and the one not
rotating) are apparently equivalent, and it becomes incompre-
hensible why one of them is spherical and the other one is not.
26
vol.2, art.137, p.724.
74
But the paradox vanishes as soon as we acknowledge the legit-
imacy of the concept of “acceleration relative to space”¿¿ 27.
Mach’s ideas profoundly influenced Einstein’s views on gra-
vity during the construction of general relativity theory. Ein-
stein wrote in one of his works: “Mach’s principle: the G-field
is fully determined by the masses of bodies.” But this statement
turns out to be not valid in GRT, since there exist solutions
in the absence of matter, also. Attempts at eliminating this
circumstance by introduction of the λ-term did not lead to
the desired result. It turned out to be that equations with the
λ-term also have solutions differing from zero in the absence
of matter. We see that Einstein attached a totally different
meaning to the concept of “Mach’s principle”. But within
such an interpretation, also, no place was found in GRT for
Mach’s principle.
Is there any place in RTG for Mach’s principle as formu-
lated by Einstein? Unlike GRT, in this theory spacelike sur-
faces are present throughout the entire space (global Cauchy
surfaces), owing to the causality principle. And if no matter
is present on one of such surfaces, then the requirement of en-
ergodominance imposed on the tensor of matter will result in
matter always being absent [26]. It will be shown in section
10 that a gravitational field cannot arise without matter.
Only solutions of the set of inhomogeneous gravitational
equations have a physical sense, when matter exists in some
part of space or throughout the entire space. This means that
the gravitational field and the effective Riemannian space in
the actual Universe could not arise without the matter that
produced them. Solution of the equations for the metric of
effective Riemannian space in the absence of matter can, for
example, be considered a limit case of the solution obtained for
a homogeneous and isotropic distribution of matter in space,
as the density of matter tends subsequently toward zero. We
27
V.A.Fock. Theory of space, time and gravity. M.:Gostekhizdat,
1961, p.499.
75
see that Mach’s principle, even as formulated by Einstein, is
realized in relativistic theory of gravity.
There exists, however, an essential difference between the
understanding of the G-field in our theory and in GRT. Ein-
stein understood the G-field to be the Riemannian metric,
while in our opinion the gravitational field is a physical field.
Such a field is present in the Riemannian metric together with
the plane metric, and therefore the metric does not vanish in
the absence of matter and of the gravitational field, but re-
mains the metric of Minkowski space.
In the literature there also exist other formulations of Mach’s
principle, differing in meaning from the ideas of both Mach and
Einstein, but since, in our opinion, these formulations are not
sufficiently clear, we have not dealt with them. Since gravita-
tional forces in RTG are due to a physical field of the Faraday–
Maxwell type, any common unique essence of the forces of
inertia and of gravity is, in principle, out of the question.
Sometimes the essence of Mach’s principle is seen to con-
sist in that the forces of inertia are determined, allegedly in
compliance with this principle, by interaction with matter in
the Universe. From a field standpoint such a principle cannot
exist in Nature. The point is that, although inertial reference
systems, as we have seen above, are related to the distribution
of matter in the Universe, forces of inertia do not result from
the interaction with matter in the Universe, because any influ-
ence of matter can only be exerted via physical fields, but this
means that the forces produced by these fields, owing to their
vector nature, cannot be made equal to zero by a choice of ref-
erence system. Thus, forces of inertia are directly determined
not by physical fields, but by a rigorously defined structure of
geometry and by the choice of reference system.
The pseudo-Euclidean geometry of space–time, which re-
flects dynamic properties common to all forms of matter, on
the one hand confirmed the hypothetical existence of inertial
reference systems, and on the other hand revealed that forces
of inertia, arising under an appropriate choice of reference sys-
76
tem, are expressed via the Christoffel symbols of Minkowski
space. Therefore, they are independent of the nature of the
body. All this became clear when it was shown that special
relativity theory is applicable not only in inertial reference
systems, but also in non-inertial (accelerated) systems.
This made it possible to provide in Ref. [7] a more general
formulation of the relativity principle: “Whatever physical ref-
erence system (inertial or non-inertial we choose, it is always
possible to indicate an infinite set of such other reference sys-
tems, in which all physical phenomena proceed like in the ini-
tial reference system, so we have and can have no experimental
for determining precisely in which reference system of this in-
finite set we happen to be.” Mathematically this is expressed
as follows: consider the interval in a certain reference system
of Minkowski space to be
dσ 2 = γµν (x)dxµ dxν ,
then there exists another reference system x! :
x!ν = f ν (x),
in which the interval assumes the form
dσ 2 = γµν (x! )dx!µ dx!ν ,
where the metric coefficients γµν have the same functional form
as in the initial reference system. In this case it is said that the
metric is form-invariant relative to such transformations,
and all physical equations are also form-invariant, i.e.
they have the same form both in the primed and in the not
primed reference systems. The transformations of coordinates
that leave the metric form-invariant form a group. In the case
of Galilean coordinates in an inertial reference system these
are the usual Lorentz transformations.
In RTG there exists an essential difference between the
forces of inertia and the forces of gravity consisting in that
77
as the distance from bodies increases, the gravitational field
becomes weaker, while the forces of inertia may become in-
definitely large, depending on the choice of reference system.
And only in an inertial reference system are they equal to zero.
Therefore, it is a mistake to consider forces of inertia insep-
arable from forces of gravity. In everyday life the difference
between them is nearly obvious.
The construction of RTG has permitted to establish the
relationship between an inertial reference system and the dis-
tribution of matter in the Universe and, thus, to understand
more profoundly the nature of forces of inertial and their differ-
ence from material forces. In our theory forces of inertia are
assigned the same role as the one they assume in any other
field theories.
8. Post-Newtonian approximation
The post-Newtonian approximation is quite sufficient for study-
ing gravitational effects in the Solar system. In this section
we shall construct this approximation. Technically, our con-
struction takes advantage of many results previously obtained
by V.A.Fock [24], and it turns out to be possible to further
simplify the method of deriving the post-Newtonian approxi-
mation.
We shall write the main equations of theory in the form
(see Appendix D)
√
γ̃ αβ Dα Dβ Φ̃-λ + m2 −γ Φ̃-λ = −16πg(TM-λ
+ τg-λ ), (8.1)
Dλ Φ̃-λ = 0 . (8.2)
-λ
where TM is the energy-momentum tensor of matter; τg-λ is
the energy-momentum tensor of the gravitational field.
The expression for the energy-momentum tensor of the
gravitational field can be represented in the form
1 1 1
− 16πgτg-λ = (g̃ -αg̃ λβ − g̃ -λg̃ αβ )(g̃νσ g̃τ µ − g̃τ σ g̃νµ )Dα Φ̃τ σ ×
2 2 2
× Dβ Φ̃ + g̃ g̃τ σ Dα Φ̃ Dβ Φ̃ − g̃ g̃τ σ Dα Φ̃λσ Dβ Φ̃ατ −
µν αβ -τ λσ -β
1
− g̃ λα g̃τ σ Dα Φ̃βσ Dβ Φ̃-τ + g̃ -λg̃τ σ Dα Φ̃σβ Dβ Φ̃ατ +
2
+ Dα Φ̃ Dβ Φ̃ − Φ̃ Dα Dβ Φ̃-λ −
-β λα αβ
√ √ 1
− m2 ( −gg̃ -λ − −γ Φ̃-λ + g̃ -α g̃ λβ γαβ − g̃ -λg̃ αβ γαβ ) . (8.3)
2
This expression is written in an arbitrary reference system in
Minkowski space. We shall further perform all computations
in the Galilean coordinates of the inertial reference system,
γµν = (1, −1, −1, −1) . (8.4)
In constructing the series of perturbation theory it is nat-
ural to apply as a small parameter such a quantity % that
v ∼ %, U ∼ %2 , Π ∼ %2 , p ∼ %2 . (8.5)
79
Here U is the Newtonian potential of the gravitational field;
Π is the specific internal energy of the body considered; p is
the specific pressure.
For the Solar system the parameter %2 is of the order of
%2 ∼ 10−6 . (8.6)
We shall use the expansions of the components of the den-

sity of the tensor:
(2) (4)
00 00 00
g̃ = 1+ Φ̃ + Φ̃ + ..., (8.7)
(3) (5)
0i 0i 0i
g̃ = Φ̃ + Φ̃ + ..., (8.8)
(2) (4)
g̃ ik = γ̃ ik + Φ̃ ik + Φ̃ ik + ... (8.9)
We shall adopt the ideal fluid model of matter, the energy-
momentum tensor of which has the form
T -λ = [p + ρ(1 + Π)]u- uλ − pg -λ, (8.10)
where ρ is the invariant density of an ideal fluid;

p is the specific isotropic pressure;
uλ is the velocity four-vector.
We shall now write the expansion in the small parameter
% for the energy-momentum tensor of matter:
(0) (2)
00
TM = T 00 + T 00 + ..., (8.11)
(1) (3)
0i
TM = T 0i + T 0i + ..., (8.12)
(2) (4)
ik
TM = T ik + T ik + ... (8.13)
In the Newtonian approximation, i.e., when we neglect the
forces of gravity, we have for the four-vector the following:
u0 = 1 + 0(%2 ), ui = v i (1 + 0(%2 )) . (8.14)

80
Substituting these expressions into (8.10) we find
(0) (1) (0)
00
T = ρ, T 0i = ρv i , T ik = 0 . (8.15)
In this approximation, on the basis of (5.7), we have
∂0 ρ + ∂i (ρv i ) = 0 . (8.16)
Hence it can be seen that in the Newtonian approximation

the total inert mass of a body is a conserved quantity:
#
M= ρd3 x . (8.17)
On the basis of equations (8.1) we have in the Newtonian

approximation:
(2)
00
∆ Φ̃ = −16πρ , (8.18)
(3)
∆ Φ̃ = −16πρv i ,
0i
(8.19)
(2)
ik
∆ Φ̃ = 0 . (8.20)
In an inertial reference system the mass of the graviton,
owing to its smallness, plays an insignificant role for effects in
the Solar system, and therefore in deriving equations (8.18) –
(8.20) we did not take it into account. But even in this case
its influence is manifested in that equations (8.2) have to exist
together with the set of equations (8.1). Such equations in
Galilean coordinates were also applied in V.A.Fock’s theory
of gravity, but unlike the case of RTG, they did not follow
from the least action principle, so it was not clear why pre-
cisely they had to be applied, instead of some other equations.
V.A. Fock chose them as coordinate conditions and applied
them in studying island systems. In RTG these equations
arise from the least action principle, and for this reason they
are universal. It is precisely owing to equations (8.2) that we
obtain a complete set of equations for determining physical
81
quantities. It must be noted that in the general case of a non-
inertial reference system or in the case of strong gravitational
fields the term with the graviton mass m can no longer be
dropped. Thus, for example, even for a static body in the
region close to the Schwarzschild sphere the influence of the
graviton mass is very significant, so it can no longer be ne-
glected.
The solution of equations (8.18) – (8.20) has the form
(2)
ρ
#
00
Φ̃ = 4U, U= d3 x! , (8.21)
|x − x! |
(3)
ρv i
#
Φ̃ = −4V i ,
0i
Vi =− d3 x! , (8.22)
|x − x! |
(2)
ik
Φ̃ = 0 . (8.23)
On the basis of equations (8.2) we have
(2) (3)
00 0i
∂0 Φ̃ + ∂i Φ̃ = 0 . (8.24)
Substituting (8.21) and (8.22) we find
∂0 U − ∂i V i = 0 . (8.25)
Hence, it is evident that differentiation of the potential U with

respect to time increases the order of smallness in %. We shall
take advantage of this circumstance in calculating the energy-
momentum tensor of the gravitational field, τg-λ . We note that
equation (8.25) is satisfied identically by virtue of equations
(8.16).
From (8.22) and (8.23) it follows that of all the density
(2)
components of the tensor Φ̃-λ only the component Φ̃ 00 , deter-
mined by expression (8.21), is seen to remain in the second
82
approximation. Precisely this circumstance significantly sim-
plifies the method of finding the post-Newtonian approxima-
tion, when at each stage of the construction we make use of
the densities of tensor quantities.
Making use of (8.21) – (8.23) with a precision up to the
second order inclusively we obtain
√ √
−gg 00 = 1 + 4U, −gg 11 =
√ √
= −gg 22 = −gg 33 = −1 . (8.26)
Hence, we have
−g = 1 + 4U , (8.26a)
consequently,
g00 = 1 − 2U, g11 = g22 = g33 = −(1 + 2U) . (8.27)
We see from (8.26) that in the Newtonian approximation,

when it suffices to consider only one component of the density
of the tensor of matter, T 00 , the gravitational field is described,
as it was expected, by only a sole component Φ̃00 , while the
metric tensor gµν has in this approximation, also, several com-
ponents, in accordance with (8.27). Working with the field
components Φ̃µν , instead of the metric tensor gµν , significantly
simplifies the entire computational process of constructing the
post-Newtonian approximation. Precisely for this reason, in-
troduction of the density of the tensor of the gravitational field
Φ̃µν is important not only from a general theoretical point of
view, but from a practical standpoint, also. Thus, the metric
tensor of the effective Riemannian space is
g00 = 1 − 2U, g0i = 4γik V k , gik = γik (1 + 2U) . (8.28)
From expression (8.21) for U it follows that the inert mass

(8.17) is equal to the active gravitational mass. In RTG, as we
have seen, this equality arose because the energy-momentum
tensor is the source of the gravitational field.
83
We shall now proceed to construct the next approximation
for the component of the metric tensor g00 . For this purpose we
shall find the contribution of the energy-momentum tensor of
the gravitational field. Since in expression (8.3) it is necessary
(2)
under the derivative sign to take into account only Φ̃ 00 , the
first term in (8.3) will give a contribution equal to
2(grad U)2 , (8.29)
while the second term contributes
− 16(grad U)2 . (8.30)
The contribution from all the remaining terms in this approx-
imation will be zero. Discarded are also the terms with time
derivatives of the potential U, since, owing to (8.25), they are
also all of a higher order of smallness in %. From (8.29) and
(8.30) we have
− 16πgτg00 = −14(grad U)2 . (8.31)
Making use of (8.31), equation (8.1) for component Φ̃00 in this

approximation assumes the form
(4)
00
∆ Φ̃ = 16πgT 00 + 14(grad U)2 + 4∂02 U . (8.32)
Since on the basis of (8.28) the interval equals the following in
the second order in %:
1
ds = dt(1 − U + vi v i) , (8.33)
2
we hence obtain
dt 1
u0 = = 1 + U − vi v i . (8.34)
ds 2
Substituting this expression into (8.10) we find
(2)
00
T = ρ[2U + Π − vi v i ] . (8.35)
84
On the basis of (8.26a) and (8.35) we obtain from equations
(8.32) the following:
(4)
∆ Φ̃ 00 = −96πρU + 16πρvi v i +
+14(grad U)2 − 16πρΠ + 4∂02 U . (8.36)
Now we shall take advantage of the obvious identity

1
(grad U)2 = ∆U 2 − U∆U . (8.37)
2
But, since
∆U = −4πρ , (8.38)
then equation (8.36), upon utilization of (8.37) and (8.38),
assumes the form
(4)
∆( Φ̃ 00 − 7U 2 ) = 16πρvi v i − 40πρU − 16πρΠ + 4∂02 U . (8.39)
Hence, we have
(4)
1 U
#
00
Φ̃ = 7U 2 + 4Φ1 + 10Φ2 + 4Φ3 − ∂02 d3 x! , (8.40)
π |x − x! |
where
#
ρvi v i 3 ! #
ρU
Φ1 = − d x, , Φ2 = d3 x! ,
|x − x! | |x − x! |
(8.41)
ρΠ
#
Φ3 = !
d3 x! .
|x − x |
Thus, in the post-Newtonian approximation we find:
g̃ 00 = 1 + 4U + 7U 2 + 4Φ1 + 10Φ2 +
1 2 U
#
+ 4Φ3 − ∂0 d3 x! . (8.42)
π |x − x! |
85
We now have to find the determinant of g in the post-Newtonian
approximation. To this end we represent g̃ ik in the form:
(4)
g̃ ik = γ̃ ik + Φ̃ ik . (8.43)
It must be especially underlined that calculation of the deter-

minant of g is most readily performed if one takes advantage
for this purpose of the tensor density g̃ µν and takes into ac-
count that
g = det(g̃ µν ) = det(gµν ) . (8.44)
From (8.42) and (8.43) we find
√ 3
−g = 1 + 2U + U 2 + 2Φ1 + 5Φ2 +
2
1 1 2 U
#
+2Φ3 − Φ − ∂0 !
d3 x! . (8.45)
2 2π |x − x |
Here
(4) (4) (4)
Φ = Φ̃ 11 + Φ̃ 22 + Φ̃ 33 . (8.46)
Since in the considered approximation g00 g 00 = 1, from ex-
pressions (8.42) and (8.45) we obtain
5
g00 = 1 − 2U + U 2 − 2Φ1 − 5Φ2 −
2
1 1 2 U
#
−2Φ3 − Φ + ∂0 d3 x! . (8.47)
2 2π |x − x! |
For determining g00 we need to calculate the quantity Φ. Since

Φ, in accordance with (8.46), was derived by summation, it is
possible to make use of equation (8.1) and by summation to
obtain directly equations for function Φ.
From expression (8.3) by summation we derive from the
first term the following expression:
− 16πgτgii = −2(grad U)2 . (8.48)
86
All the remaining terms present in expression (8.3) give no
contribution in this approximation. With the aid of expression
(8.10) for the energy-momentum tensor we find
(2)
− 16πg T̃ = −16πρvi v i + 48πp .
ii
(8.49)
Taking into account (8.48) and (8.49), we write the equation

for Φ as follows:
∆Φ = 16πρvi v i − 48πp + 2(grad U)2 . (8.50)
Taking advantage of identity (8.37) and of equation (8.38) we

obtain
∆(Φ − U 2 ) = 16πρvi v i + 8πρU − 48πp . (8.51)
Hence, we find
Φ = U 2 + 4Φ1 − 2Φ2 + 12Φ4 , (8.52)
where #
p
Φ4 = d3 x! .
|x − x! |
Substituting expression (8.52) into (8.47) we have
g00 = 1 − 2U + 2U 2 − 4Φ1 − 4Φ2 −

1 #
U
−2Φ3 − 6Φ4 + ∂02 d3 x! . (8.53)
2π |x − x! |
Making use of the identity
1 U
# #
!
d3 x! = − ρ|x − x! |d3x! ,
2π |x − x |
we write expression (8.53) as
g00 = 1 − 2U + 2U 2 − 4Φ1 − 4Φ2 −

#
− 2Φ3 − 6Φ4 − ∂02 ρ|x − x! |d3 x! . (8.54)
87
The solutions (8.54) and (8.28) are calculated in an iner-
tial reference system in Galilean coordinates. The effective
Riemannian metric that arises is due to the presence of the
gravitational field, while the forces of inertia are totally ex-
cluded. It is quite obvious that these solutions retain their
functional form in the Galilean coordinates of any inertial ref-
erence system. Since all physical quantities are independent
of transformations of the time variable, then if the following
transformation is applied:
x!0 = x0 + η 0 (x), x!i = xi , (8.55)
the metric coefficients will change as follows:
!
g00 = g00 − 2∂0 η 0 , g0i
!
= g0i − ∂i η 0 , gik
!
= gik . (8.56)
It must be noted that transformation (8.55) does not take us
beyond the inertial reference system, since such a transfor-
mation is nothing but another choice of clock. All physically
measurable quantities are independent of this choice.
Assuming function η 0 to be
1
#
η 0 = − ∂0 ρ|x − x! |d3 x! , (8.57)
2
and taking into account the identity
1
∂i η 0 = (γik V k − Ni ),
#2 k (8.58)
ρv (xk − x!k )(xi − x!i ) 3 !
Ni = d x
|x − x! |3
upon substitution into (8.56) of expressions (8.28) for g0i and
gik and, also, of expression (8.54) for g00 , and taking into ac-
count(8.57) and (8.58), we find the metric coefficients of effec-
tive Riemannian space in the so-called ”canonical form”:
g00 = 1 − 2U + 2U 2 − 4Φ1 − 4Φ2 − 2Φ3 − 6Φ4 ,
7 1
g0i = γik V k + Ni , (8.59)
2 2
gik = γik (1 + 2U) .
88
These expressions coincide precisely with the formulae that
are obtained on the basis of GRT. The difference only consists
in that here they follow exactly from RTG, while for deriving
them from GRT equations it is necessary to apply additional
assumptions, that do not follow from theory, i.e. it is necessary
to go beyond the limits of GRT. But we shall specially deal
with this issue.
In the case of a static spherically symmetric body the
post-Newtonian approximation at a distance from the body
assumes, in accordance with (8.59), the form
'2
2MG MG
&
g00 =1− +2 , g0i = 0,
r r
(8.59a)
2MG
& ' #
3
gik = γik 1 + , M = ρ(x)d x.
r
On the basis of expressions (8.59) the post-Newtonian No-
rdtwedt–Will parameters in RTG assume the following values:
γ = 1, β = 1, α1 = α2 = α3 = ξ1 = ξ2 = ξ3 = ξ4 = ξW = 0.
We have calculated the metric coefficients (8.59) in RTG in an

inertial reference system. We shall now present the expressions
for the components of the energy-momentum tensor of matter
in the next approximation, as compared with (8.15). Taking
into account expression (8.34) for u0 and, also, that
dxi 1
ui = = v i (1 + U − vk v k ) , (8.60)
ds 2
we find from formula (8.10)
(3)
0i
T = ρv i (2U + Π − vk v k ) + pv i , (8.61)
(2)
ik
T = ρv i v k − pγ ik . (8.62)
(2)
The component T 00 is determined by expression (8.35).
On the basis of expression (8.59), making use of the equations
89
for the geodesic line, it is possible to calculate all effects in
the Solar system. When gravitational effects in the Solar sys-
tem are calculated in GRT on the basis of the post-Newtonian
approximation, the results obtained are correct, and no ambi-
guity is present in the description of the effects. At the same
time, if the exact solutions of GRT are applied, ambiguity
arises in the description of the effects.
In conclusion we shall deal in somewhat greater detail with
the comparison of RTG and GRT in analyzing effects occurring
in a weak gravitational field. The set of equations (8.1) and
(8.2) together with the equation of state determines all the
physical quantities of one or another gravitational problem.
All the calculations performed above in the post-Newtonian
approximation were made in an inertial reference system. In
GRT there in principle exists no inertial reference system. In
this connection A. Einstein wrote: “The starting point of the-
ory consists in the assertion that there exists no state of motion
physically singled out, i.e. not only velocity, but acceleration,
also, have no absolute meaning” 28. But if no inertial reference
system exists, to which reference system must one consider
calculations performed within GRT to pertain?
In calculating gravitational effects V. A. Fock made use
of the harmonicity conditions in Cartesian coordinates. He
called them coordinate conditions. Thus, in a work published
in 1939 [24] he wrote: “In solving Einstein’s equations we
took advantage of a reference system, which we have termed
harmonic, but which merits being called inertial.” Further in
the same article he noted: “It seems to us that the possibil-
ity of introducing in general relativity theory a definite inertial
reference system in an unambiguous manner is noteworthy.”
And, finally, in Ref. [25] he wrote: “The relativity principle
expressed by the Lorentz transformations is possible in inho-
28
vol.2, art.92, p.264.
90
mogeneous space, also, while a general relativity principle is
not possible.”
All these statements of V. A. Fock were due to his aspira-
tion to clarify the essence of GRT, freeing it of general rela-
tivity devoid of any physical meaning. However, V. A. Fock,
here, actually went beyond the limits of GRT. Precisely owing
to this fact he arrived at the striking conclusion on the vali-
dity of the relativity principle in inhomogeneous space, also.
If one remains within Riemannian space, and no other space
exists in GRT, then this assertion contradicts the correct con-
clusion made by V. A. Fock “that in general relativity theory
there, generally speaking, exists no relativity.” [25]. But to
realize his goal it is necessary to introduce the concept of a
gravitational field in Minkowski space. Where did V. A. Fock
go beyond GRT? In applying the conditions of harmonicity he
actually considered Cartesian coordinates:
∂g̃ µν
=0, (8.63)
∂xµ
where xµ are Cartesian coordinates. In Cartesian coordinates
γ(x) = det γµν = −1. Therefore, in accordance with the tensor
law of transformations we have
∂xµ ∂xν g̃ αβ (y)
g̃ µν (x) = · ·- . (8.64)
∂y α ∂y β −γ(y)
We shall write equations (8.63) in the form
∂y τ ∂g̃ µν (x)
∂µ g̃ µν (x) = · . (8.65)
∂xµ ∂y τ
For further calculations we present the formulae
 
∂  1  = −√
1 λ ν ∂ 2 xσ ∂y ν
γ , γ = · . (8.66)
−γ τ λ αβ ∂y α ∂y β ∂xσ
-
∂y τ −γ(y)
91
Upon substitution of (8.64) into (8.65) and taking into account
(8.66) we obtain
1 ∂xν ∂g̃ ασ (y)
∂µ g̃ µν (x) = √ · +
−γ ∂y σ ∂y α
1 ∂ 2 xν
+ √ g̃ αβ (y) α β = 0 . (8.67)
−γ ∂y ∂y
We shall write the multiplier of the second term in the form
∂ 2 xν ∂xν ∂y σ ∂ 2 xτ ∂xν σ
= · · = ·γ .
∂y α ∂y β ∂y σ ∂xτ ∂y α ∂y β ∂y σ αβ
Substituting this expression into the preceding one we find
∂xν ∂g̃ ασ (y)
$ %
µν 1 σ
∂µ g̃ (x) = √ · σ + γαβ (y)g̃ αβ (y) = 0 ,
−γ ∂y ∂y α
i.e. we have
µν 1 ∂xν
∂µ g̃ (x) = √ · σ Dµ g̃ µσ (y) = 0 . (8.68)
−γ ∂y
Thus, we have established that the density of the tensor
g̃ µσ (y) in arbitrary coordinates automatically satisfies the gen-
eral covariant equation
Dλ g̃ λσ = 0 ,
if the initial condition of harmonicity (8.63) is written in Carte-
sian coordinates. But this means that the harmonicity con-
dition is not a coordinate condition, but a field equation in
Minkowski space. Thus, application of the condition of har-
monicity in Cartesian coordinates is not an innocent opera-
tion, but it implies going beyond the framework of GRT by
introduction of Minkowski space.
The obtained equation coincides with equation (5.20) of
RTG. In RTG it follows from the least action principle. Per-
forming transformation from coordinates y to coordinates z
we obtain (see Appendix (E.12))
y λ = −γαβ
λ
(y)g αβ (y),
92
where denotes the operator
$ %
1 ∂ νσ ∂
=- · ν g̃ .
−g(z) ∂z ∂z σ
Therefore, when V. A. Fock wrote down the harmonicity con-

ditions in the form
y λ = 0,
he actually dealt with Cartesian coordinates, for which
λ
γαβ (y) = 0, i.e. with Minkowski space in Galilean coordi-
nates. In choosing harmonic coordinates in the form of con-
ditions (8.63) V. A. Fock actually made use of Minkowski
space in Galilean coordinates, while equations (8.63) played
the part of field equations, instead of coordinate conditions.
But why it was necessary to add to the Hilbert–Einstein equa-
tions precisely equations (8.63) in Galilean coordinates, in-
stead of some others, in order to obtain the complete set of
gravity equations within V.A.Fock’s approach, remained un-
clear. Here, V.A.Fock was most likely guided by physical intu-
ition, and also by the mathematical simplification that arose
in the course of calculations.
Did V. A. Fock attempt to consider the gravitational field
in Minkowski space? No, he was far from this idea, and in this
connection he wrote [24]: “We recall it here only in connection
with the sometimes observed tendency (certainly not shared by
us) to pack the theory of gravity into the framework of Eu-
clidean space.” As we have seen, application of the conditions
of harmonicity in Cartesian coordinates makes us go beyond
the framework of GRT. But this means that the set of equa-
tions of gravity, studied by V. A. Fock, differs from the set of
equations of GRT, i.e. V. A. Fock’s theory of gravity based
on the conditions of harmonicity in Cartesian coordinates and
Einstein’s GRT are different theories. V. A. Fock’s approach
turns out to be closer to the ideas of RTG. Everything that
V. A. Fock attempted to introduce in the theory of gravity (in-
ertial reference systems, acceleration relative to space) is fully
93
inherent in RTG, but this is achieved by consideration of the
gravitational field, like all other physical fields, in Minkowski
space. Here, all the geometric characteristics of Riemannian
space are now field quantities in Minkowski space.
In analyzing gravitational effects in the Solar sys-
tem V.A. Fock actually made use of Minkowski space, since
he referred all the calculated gravitational effects to an iner-
tial reference system. Precisely this circumstance permitted
him to obtain correct expressions for the effects. Thus, for ex-
ample, he wrote [25]: “How should one define a straight line:
as a light ray or as a straight line in that Euclidean space in
which the harmonic coordinates x1 , x2 , x3 serve as Cartesian
coordinates? We think the second definition to be the only cor-
rect one. We actually made use of it in saying that a light ray
has the shape of a hyperbola in the vicinity of the Sun”, and
further on, “the argument that a straight line, like a ray of
light, is more directly observable, but it has no sense: in the
definitions it is not the direct observability that is decisive, but
the correspondence to Nature, even though this correspondence
may be established by indirect reasoning.”
In RTG gravitational effects are determined unambigu-
ously, because in accordance with equations (8.1) and (8.2)
written in the Galilean coordinates of an inertial reference sys-
tem, the motion of light or of a test body, when the gravita-
tional field is switched off, indeed proceeds along a straight
line, which is a geodesic line in Minkowski space. It is abso-
lutely clear that in a non-inertial reference system the geodesic
line in Minkowski space will no longer be a straight line. But
this means that in RTG, in an non-inertial reference system,
for revealing a gravitational effect motion in effective Rieman-
nian space must be compared precisely with the geodesic mo-
tion of the accelerated reference system.
In calculating gravity effects in the Solar system, when the
influence of the graviton mass can be neglected, only the RTG
set of equations (8.1) and (8.2) in Galilean coordinates coin-
cides with the set of equations dealt with by V. A. Fock in
94
harmonic (Cartesian) coordinates. If one remains within the
framework of GRT, then in any other, for instance non-inertial,
reference system they differ essentially. This takes place be-
cause the covariance of V. A. Fock’s set of equations is not
general, unlike the set of RTG equations. V. A. Fock obtained
the complete set of gravitational equations (for island systems)
by adding the harmonicity conditions to the Hilbert-Einstein
equations. But why must precisely the harmonicity condi-
tions be added, instead of some other conditions, remained
unclear. In accordance with RTG, the complete set of gravi-
tational equations (8.1) and (8.2) arises from the least action
principle. Hence it becomes clear, why conditions (8.2), which
in Cartesian coordinates coincide with the harmonicity condi-
tions, arise, instead of some other conditions. But these equa-
tions become universal, valid not only for island systems. But
if V. A. Fock had realized that in applying the harmonicity
conditions he actually had to deal with Cartesian coordinates
of Minkowski space, he would have readily obtained expression
(8.68). As we already noted earlier, the harmonicity conditions
in Cartesian coordinates successfully applied by V. A. Fock
took him beyond the framework of Einstein’s GRT. This fact
was noted in 1957 by L. Infeld who wrote:“Thus, for Fock
the choice of the harmonicity coordinate condition becomes a
certain fundamental law of Nature, which alters the very char-
acter of Einstein’s general relativity theory and transforms it
into a theory of the gravitational field, valid only in inertial
reference systems” 29 .
If one remains within the framework of GRT, then it is ab-
solutely incomprehensible, from the point of view of physics,
why it is necessary to choose the harmonicity conditions, in-
stead of any other conditions. While in RTG, owing to the
existence of the graviton mass, these conditions arise as a con-
sequence of the validity of the equations for matter [see (5.7)
29
L.Infeld. Most recent problems in gravity. M.:Foreign literature
95
and (5.17)], i.e. they follow from the least action principle, and
they therefore have universal significance. However, in GRT
similar expressions for the post-Newtonian approximation are,
nevertheless, obtained without application of the harmonici-
ty conditions in Cartesian coordinates. Why is this so? The
reason consists in that Minkowski space in Galilean coordi-
nates is once again introduced and that the gravitational field
is actually considered as a physical field in this space.
The metric of Minkowski space in Galilean coordinates is
taken as the zero order approximation for the Riemannian
metric. It is complemented with various potentials with arbi-
trary post-Newtonian parameters, each of which decreases like
0( 1r ). In this way the arbitrariness contained in GRT is dis-
carded. Substitution of the Riemannian metric gµν in this form
into the Hilbert-Einstein equation permits one to determine
the values of the post-Newtonian parameters, and we again
arrive at the same post-Newtonian approximation. Precisely
here gravity is considered to be a physical field in Minkowski
space, the behaviour of which is described by the introduced
gravitational potentials. Such a requirement imposed on the
character of the metric of Riemannian space does not follow
from GRT, since in the general case the asymptotics of the
metric is quite arbitrary and even depends on the choice of
the three-dimensional space coordinates. Therefore it is im-
possible to impose physical conditions on the metric. But if
it is effective and its arising is due to the physical field, then
the physical conditions are imposed on the metric in a natural
manner.
In RTG the gravitational equations (5.19) and (5.20) are
generally covariant, but not form-invariant with respect to ar-
bitrary transformations. They are form-invariant relative to
the Lorentz transformations. But this means that in Lorentz
coordinates, in case the solution G(x) exists for the tensor of
matter Tµν (x), there exists, in the new Lorentz coordinates x! ,
the solution G! (x! ) for the tensor of matter Tµν
!
(x! ), and, con-
96
sequently, in the coordinates x the solution G! (x) is possible
!
only for the tensor of matter Tµν (x).
In RTG a unique correspondence is established between the
Riemannian metric and the Minkowski metric, which permits
one to compare motion under the influence of the gravitational
field and in its absence, when calculation is performed of the
gravitational effect. When the gravitational field is switched
off in RTG the Riemann tensor turns to zero, and at the
same time transition occurs from Riemannian metric to the
Minkowski metric, previously chosen in formulating the phys-
ical problem. This is precisely what provides for the equiva-
lence principle to be satisfied in RTG.
For calculation of the gravitational effect it is necessary to
compare motion in Riemannian space with motion in absence
of the gravitational field. This is precisely how the gravita-
tional effect is determined. If in GRT one refers the set of
solutions for gµν to a certain inertial reference system, then it
is quite obvious that one will obtain a whole set of various val-
ues for the gravitational effect. Which one of them should be
chosen? Since the Hilbert-Einstein equations do not contain
the metric of Minkowski space, it is impossible to satisfy the
equivalence principle, because it is impossible to determine, in
which (inertial or non-inertial) reference system one happens
to be, when the gravitational field is switched off.
To conclude this section we note that the post-Newtonian
approximation (8.59) satisfies the causality principle (6.11).
9. On the equality of inert and
gravitational masses
Owing to the density of the energy-momentum tensor being
the source of the gravitational field, the inert and gravitational
masses were shown in section 8 to be equal. In this section we
shall show that the field approach to gravity permits obtaining
in a trivial manner the metric of effective Riemannian space in
the first approximation in the gravitational constant G. This
is especially simple to establish on the basis of equations (2.2).
In the case of a spherically symmetric static body, equations
(2.2) have the following form in the Galilean coordinates of an
inertial reference system:
∆Φ̃00 − m2 Φ̃00 = −16πt00 , (9.1)
∆Φ̃0i − m2 Φ̃0i = 0, ∆Φ̃ik − m2 Φ̃ik = 0, i, k = 1, 2, 3. (9.2)
For a static body the sole component t00 differs from zero.
From equations (9.2) we have
Φ̃0i = 0, Φ̃ik = 0 . (9.3)
Far away from the body, from equation (9.1) we find
4M −mr
#
Φ̃00 + e , M= t00 d3 x , (9.4)
r
M is the inert mass of the body, that creates the gravitational
field. In the Solar system the exponential factor can be ne-
glected, owing to the quantity mr being small.
4M
Φ̃00 + . (9.5)
r
We shall now find the components of the density of the
metric tensor of effective Riemannian space, g̃ µν . On the basis
of (2.6) we have
√
g̃ µν = γ̃ µν + Φ̃µν , g̃ µν = −gg µν . (9.6)
98
Hence, taking into account (9.3) and (9.5), we obtain the fol-
lowing g̃ µν components, that differ from zero:
4M
g̃ 00 = 1 + , g̃ 11 = g̃ 22 = g̃ 33 = −1. (9.7)
r
They satisfy equation (2.3) exactly. On the basis of (9.7) we
find √
−g √
g00 = 4M , g11 = g22 = g33 = − −g. (9.8)
1+ r
4M
& '
00 11 22 33
− g = − g̃ g̃ g̃ g̃ = 1 + . (9.9)
r
Substituting the expressions for g into formulae (9.8) we obtain
2M 2M
& ' & '
g00 + 1 − , g11 = g22 = g33 = − 1 + . (9.10)
r r
It must be especially underlined that at the place, where
in accordance with Newton’s law of gravity there should be
an active gravitational mass, there appears the inert mass M.
Thus, the equality of the inert and active gravitational masses
is a direct consequence of the density of the energy-momentum
tensor being the source of the gravitational field. So the reason
that the inert and gravitational masses are equal is not the lo-
cal identity of the forces of inertia and of gravity (this actually
does not occur in GRT), but the universality of the conserved
source of the gravitational field, of the energy-momentum ten-
sor of matter.
The interval in effective Riemannian space has the form
2M 2M
& ' & '
ds2 = 1 − dt2 − 1 + (dx2 + dy 2 + dr 2) . (9.11)
r r
Classical effects of gravity, such as the gravitational red shift
of spectral lines, the deviation of a light ray by the Sun, the
time delay of a radiosignal, the precession of a gyroscope on
the Earth’s orbit, are fully described by this interval.
99
From expression (9.10) it is evident that the forces of gra-
vity are attractive, since the quantity M, being an inert mass,
is always positive. As to GRT, in accordance with this theory
it is not possible to prove the equality of inert and active grav-
itational masses. A detailed analysis of this issue is presented
in joint works performed with prof. V. I. Denisov. This cir-
cumstance is dealt with in detail in the monograph [10]. The
essence of the issue consists in that the expression for inert
mass, determined from the pseudotensor of the gravitational
field, depends on the choice of the three-dimensional coordi-
nates, which is physically inadmissible. Precisely by a simple
choice of three-dimensional space coordinates (which is always
permitted) one can show that in GRT inert mass is not equal
to active gravitational mass. Since the equality of physically
measurable quantities in GRT depends on the choice of the
three-dimensional coordinates, this means that not everything
in it is alright here, also. Sometimes the opinion is voiced
that within the framework of GRT it is possible to construct
the energy-momentum tensor of the gravitational field by sub-
stitution of covariant derivatives in Minkowski space for the
ordinary derivatives in the expression for the pseudotensor.
However, here, on the one hand, it is impossible to say with
definiteness which metric in Minkowski space must be taken
for such a substitution, and, on the other hand, in Riemannian
space no global Cartesian coordinates exist, and, consequently,
no Minkowski space, so such an approach does not remove the
essential difficulty of GRT: the absence of integral conserva-
tion laws of energy-momentum and of angular momentum for
matter and gravitational field taken together.
10. Evolution of the homogeneous
and isotropic Universe
We write the equations of RTG in the form
m2 8π 1
& '
Rµν − (gµν − γµν ) = √ Tµν − gµν T , (10.1)
2 −g 2
Dµ g̃ µν = 0 . (10.2)
For convenience we have chosen the set of units
G = h̄ = c = 1. In the final expressions we shall restore the
dependence upon these constants. The density of the energy-
momentum tensor has the form
√ ν dxν
Tµν = −g[(ρ + p)Uµ Uν − gµν p], U = . (10.3)
ds
Here ρ is the density of matter, p is pressure, ds is the in-
terval in effective Riemannian space. For a homogeneous and
isotropic model of the Universe the interval of effective Rie-
mannian space ds has the general form
dr 2
( )
2 2
ds = U(t)dt −V (t) + r 2(dΘ2 + sin2 ΘdΦ2 ) . (10.4)
1 − kr 2
Here k assumes the values 1, −1, 0; k = 1 corresponds to the

closed Universe, k = −1 — to the hyperbolic Universe, and
k = 0 — to the “flat” Universe.
Since the set of RTG equations (10.1) and (10.2) together
with the equation of state is complete, then in the case of
appropriate initial conditions it can yield only a single solution
describing the development of a homogeneous and isotropic
model of the Universe. At the same time the equations of GRT
for the same model yield three well-known scenarios for the
development of the Universe. The scenario of the development
of the Universe obtained on the basis of the RTG does not
101
coincide with any of the scenarios based on GRT. We shall
follow [31].
All our analysis will be made in an inertial reference sys-
tem in spherical coordinates r, Θ, Φ. An interval in Minkowski
space, in this case, will have the form
dσ 2 = dt2 − dx2 − dy 2 − dz 2 =
= dt2 − dr 2 − r 2 (dΘ2 + sin2 ΘdΦ2 ) . (10.5)
The determinant g, composed of components gµν , equals
g = −UV 3 (1 − kr 2 )−1 r 4 sin2 Θ. (10.6)
The tensor density √

g̃ µν = −gg µν (10.7)
has, in accordance with (10.4), the following components:
2
r4 V 3
3
g̃ 00 =
3
4 · sin Θ,
U(1 − kr 2 )
-
g̃ 11 = −r 2 UV (1 − kr 2 ) · sin Θ ,
5 (10.8)
UV
g̃ 22 =− · sin Θ,
1 − kr 2
5
UV 1
g̃ 33 =− 2
· .
1 − kr sin Θ
The Christoffel symbols of Minkowski space are
1 1 1
γ22 = −r, γ33 = −r sin2 Θ, γ12
2 3
= γ13 = ,
r (10.9)
2 3
γ33 = − sin Θ cos Θ, γ23 = cot Θ .
Equation (10.2) has the form
Dµ g̃ µν = ∂µ g̃ µν + γαβ
ν
g̃ αβ = 0 . (10.10)
102
Substituting (10.9) into (10.10) we obtain
V3
$ %
∂
=0, (10.11)
∂t U
∂ 2√
(r 1 − kr 2 ) = 2r(1 − kr 2 )−1/2 . (10.12)
∂r
From equation (10.11) it follows
V = aU 1/3 ,
here a is an integration constant.

From equation (10.12) we directly find
k=0, (10.13)
i.e. the space metric is Euclidean. It must be stressed that

this conclusion for a homogeneous and isotropic Universe fol-
lows directly from equation (10.2) for the gravitational field
and does not depend on the density of matter. Thus, equa-
tion (10.2) excludes the closed and hyperbolic models of the
Universe. A homogeneous and isotropic Universe can only
be “flat” according to RTG. In other words, the well-known
problem of the Universe’s flatness does not exist within the
framework of RTG. With account of (10.11) and (10.13) the
effective Riemannian metric (10.4) assumes the form
ds2 = U(t)dt2 − aU 1/3 [dr 2 + r 2 (dΘ2 + sin2 ΘdΦ2 )] . (10.14)
If one passes to the proper time dτ

√
dτ = U dt (10.15)
and introduces the notation
R2 = U 1/3 (t) , (10.16)
the interval (10.14) assumes the form
ds2 = dτ 2 − aR2 (τ )[dx2 + dy 2 + dz 2 ] . (10.17)

103
Here and further in this section R is a scaling factor. We are
compelled to make use of the notation adopted in the litera-
ture for this quantity, in spite of the fact that in the previous
sections R stood for the scalar curvature. For the given metric
the Christoffel symbols assume the form
1
Γ122 = −r, Γ133 = −r sin2 Θ, Γ212 = Γ313 = ,
r (10.18)
2 3
Γ33 = − sin Θ cos Θ, Γ23 = cot Θ ,
dR 1 dR
Γ0ii = aR , Γi0i = , i = 1, 2, 3 . (10.19)
dτ R dτ
Making use of expression (4.13) we have
%2
3 d2 R d2 R
$
dR
R00 =− , R11 = 2a + aR ,
R dτ 2 dτ dτ 2
(10.20)
R22 = r 2 R11 , R33 = sin2 Θ · R22 , R0i = 0 ,
%2
6 d2 R
$
µν 6 dR
Rµν g =− · 2 − 2· . (10.21)
R dτ R dτ
Since g0i = 0, R0i = 0, then from equation (10.1) it follows

directly that
T0i = 0 . (10.22)
Hence on the basis of (10.3) we have
Ui = 0 . (10.23)
This means that matter is at rest in an inertial reference sys-

tem. Thus, the so-called “expansion” of the Universe, observed
by the red shift, is due to the change of the gravitational field
in time. Therefore, there exists no expansion of the Universe,
related to the motion of objects with respect to each other.
The red shift is not due to the motion of galaxies, which is
absent, but to the variation of the gravitational field in time.
Therefore, the red shift does not indicate that the galaxies
were at a time close to each other. At the same time, in
104
accordance with GRT “All versions of the Friedman model
have in common that at a certain moment of time in the past
(ten–twenty thousand million years ago) the distance between
adjacent galaxies should have been equal to zero” 30 . We, here,
have quoted S. Hawking, and below we shall establish the rea-
son for the difference between the conclusions concerning the
development of the Universe in RTG and GRT. We shall now
deal in a somewhat greater detail with the nature of the red
shift.
From (10.17) it follows that the speed of a light ray equals
dr 1
=√ .
dτ aR(τ )
Let us put the observation point at the origin of the reference

system (r = 0). Consider a light signal emitted from point
r during the time interval between τ and τ + dτ , and let its
arrival at the point r = 0 take place during the time interval
between τ0 and τ0 + dτ0 ; then, for light emitted at moment τ
and arriving at the point r = 0 at the moment τ0 we have
#τ0 √
dτ
= ar,
τ
R(τ )
similarly, for light emitted at a moment τ + dτ and arriving at

the point r = 0 at the moment τ0 + dτ0 we find
τ0#+dτ0
dτ √
= ar.
R(τ )
τ +dτ
Equating these expressions we obtain

dτ dτ0
= .
R(τ ) R(τ0 )
30
S. Hawking. From the Big Bang to Black Holes. M.:Mir, 1990,
p.46.
105
Or, passing to the light frequency, we have
R(τ0 )
ω= ω0 .
R(τ )
Hence, it is obvious that the light frequency ω at the point of
emitting is not equal to the frequency of the light ω0 at the
point of observation.
Introducing the red shift parameter z
ω − ω0
z= ,
ω0
we have
R(τ0 )
z= − 1.
R(τ )
We see that the red shift is only related to variation of the
scaling factor R(τ ), in the case of such variation there exists
no motion of matter, in accordance with (10.23). Thus, the
nature of the red shift is not related to the scattering of galax-
ies, which is absent, but to variation of the gravitational field
with time, i.e. it is related to the fact that R(τ0 ) > R(τ ).
It must be especially stressed that a given inertial reference
system is singled out by Nature itself, i.e. in the considered
theory the Mach principle is satisfied automatically.
Substituting (10.20) and (10.3) into equation (10.1), with
account of (10.23), we have
1 d2 R 4πG 3p 1
& ' & '
2
=− ρ + 2 − 2ω 1 − 6 , (10.24)
R dτ 3 c R
%2
4
$ $ %
1 dR 8πG ω 3R 6
= ρ− 6 1− + 2R . (10.25)
R dτ 3 R a
where %2
mc2
$
1
ω= . (10.26)
12 h̄
From (10.24) it is seen that for small values of the scaling
factor R there arises an initial acceleration owing to the sec-
ond term. This is precisely what “incites” the “expansion” of
106
the Universe. The initial acceleration appears at the moment
when the density of matter stops growing in the preceding cy-
cle. From (10.25) it follows that in the region R >> 1 the
contemporary density of matter in the Universe equals
%2
mc2
$
1
ρ(τ ) = ρc (τ ) + , (10.27)
16πG h̄
where ρc (τ ) is the critical density determined by the Hubble

“constant”
3H 2 (τ ) 1 dR
ρc = , H(τ ) = · . (10.28)
8πG R dτ
Hence, the necessity for the existence of “dark” matter, which
is in accordance with modern observational data.
From equations (10.24) and (10.25) one can obtain the ex-
pression for the deceleration parameter of the Universe, q(τ ):
'2 & '2
R̈ 1 1 1 c mc
&
q=− 2
= + . (10.29)
RH 2 4 H h̄
Thus, the parameter q at present is positive, i.e. “expansion”
of the Universe has slowed down, instead of being accelerated.
The relation (10.29) makes it possible, in principle, to deter-
mine the mass of the graviton from two observable quantities,
H and q. From the causality principle (6.10), (6.11) it follows
that
R2 (R4 − a) ≤ 0 . (10.30)
To satisfy the causality condition throughout the entire region
of variation of R(τ ) it is natural to set
4
a = Rmax . (10.31)
From the condition that the left-hand side of equation (10.25)

is not negative it follows that the expansion should start from
some minimum value Rmin , corresponding to the value dR dτ
= 0.
107
On the other hand, if R >> 1, the expansion should stop at
Rmax , when the density (10.27) reaches its minimum value
%2 $
mc2
$ %
1 1
ρmin = 1− 6
. (10.32)
16πG h̄ Rmax
and the process starts of compression down to Rmin .
Thus, in RTG there exists no cosmological singularity, and
the presence of the graviton mass results in the evolution of
the Universe exhibiting a cyclical character. The time required
for the Universe to expand from the maximum to its minimum
density is mainly determined by the stage at which nonrela-
tivistic matter is dominant and is
5
2 πh̄
τmax + . (10.33)
3 mc2
From the covariant conservation law, that is a consequence of
equations (5.19), (5.20)
∇µ T̃ µν + Γναβ T̃ αβ = 0
it is possible to obtain the equation
1 dR 1 dρ
=− p . (10.34)
R dτ 3(ρ + c2
) dτ
For the stage of development of the Universe dominated by
radiation
1
p = ρc2
3
from equation (10.34) we obtain the following expression for
the radiation density ρr :
A
ρr (τ ) = 4 . (10.35)
R (τ )
Here A is an integration constant. At the stage of development
of the Universe, when nonrelativistic matter is dominant and
pressure can be neglected, from equation (10.34) we find
B
ρm (τ ) = , (10.36)
R3 (τ )
108
B is an integration constant.
Consider that at a certain moment of time τ0 the radiation
density ρr (τ0 ) becomes equal to the density of matter, ρm (τ0 )
ρr (τ0 ) = ρm (τ0 ) , (10.37)
then
A = BR(τ0 ) = BR0 .
Since at later stages of the development of the Universe matter
is dominant, we have from formula (10.36) the following:
3
B = ρmin · Rmax . (10.38)
Thus,
3
ρmin R0 · Rmax
ρ + ρr = , R ≤ R0 , (10.39)
R4
Rmax 3
& '
ρ + ρm = ρmin , R ≥ R0 . (10.40)
R
According to observational data (see, for example, [33]), the
present-day density of radiation (including the three sorts of
neutrinos, which we for definiteness consider massless) and the
critical density of matter, are
ρr (τc ) = 8 · 10−34 g/cm3 , ρm (τc ) = 10−29 g/cm3 . (10.41)
The “hidden” mass must be attributed to the density of mat-

ter, ρm (τc ), for our choice of the graviton mass (m = 10−66 g)
it is close to the critical density ρc determined by the Hub-
ble “constant”. We intend matter to actually be all forms of
matter, with the exception of the gravitational field.
In accordance with formulae (10.39), (10.40) and (10.41)
we have
3
ρmin R0 · Rmax
ρr (τc ) = = 8 · 10−34 g/cm3 , (10.42)
R4 (τc )
109
$ %3
Rmax
ρm (τc ) = ρmin = 10−29 g/cm3 . (10.43)
R(τc )
Hence, we find
ρr (τc ) 1/3 1/3
R0 = 4/3
Rmax · ρmin = 3, 7 · 105 ρmin · Rmax . (10.44)
ρm (τc )
Let us introduce the notation
4 3
σ = R0 · Rmax . (10.45)
3
In accordance with (10.39), (10.44) and (10.45) we obtain
3 σ · ρmin
ρ(τ ) + ρr (τ ) = · , R ≤ R0 . (10.46)
4 R4 (τ )
Assuming radiation to be dominant at the initial stage of ex-
pansion in the hot Universe model, from equation (10.25) by
taking into account (10.31), (10.32) and (10.46), we obtain the
following:
$ %2 ( )
2 1 dR 3σ 3 1
H = =ω − 2 + − . (10.47)
R dτ 2R4 4
Rmax · R2 R6
Equation (10.47) makes it possible to determine the law of
expansion of the Universe at the initial stage. It is readily
seen that the right-hand side of equation (10.47) turns to zero
at sufficiently small values of R = Rmin . The main role, here,
is due to the first term in brackets in equation (10.25), that is
responsible for the graviton mass.
By introducing the variable x = R−2 one can readily find
approximate values for the roots of the equation
3 2 3
σx − 2 + 4 x − x3 = 0 , (10.48)
2 Rmax
which are the turning points
5
3 1 4 1 1
& ' & '
x1 = σ + 0 2 , x2,3 = ± √ +0 2 . (10.49)
2 σ 3 σ σ
110
Hence, we find the turning point
5
2
Rmin = . (10.50)
3σ
Thus, owing to the graviton mass, there exists no cosmolo-
gical singularity in RTG, and expansion of the Universe starts
from the finite non-zero value R = Rmin . On the basis of
(10.46) we obtain
3 σρmin 27
ρmax = · 4 = σ 3 · ρmin . (10.51)
4 Rmin 16
In accordance with (10.49) expression (10.47) can be written

in the form
H 2 = ω(x1 − x)(x − x2 )(x − x3 ) . (10.52)
Within the range of variation of the scaling factor
Rmin ≤ R ≤ R0 (10.53)
the expression for H 2 is significantly simplified:

3σω 2
H 2 + ωx2 (x1 − x) = 2
(R − Rmin ). (10.54)
2R6
Within this approximation equation (10.47) assumes the form
$ %2
1 dR 3σω 2
= (R − Rmin ) . (10.55)
R2 dτ 2R6
Upon integration we find
R2 √ √
τ = √ min [Z Z 2 − 1 + ln(Z + Z 2 − 1)] , (10.56)
6σω
where
Z = R/Rmin .
111
Utilizing expressions (10.50) and (10.51) we obtain
%1/2
R2
$
1 ρmin
√ min = √ . (10.57)
6σω 2 2ω ρmax
Substituting into this expression the value ρmin from (10.32)

we find 5
2
Rmin 3
√ = . (10.58)
6σω 32πGρmax
Taking into account (10.58) in (10.56) we obtain
√ √
5
3
τ= [Z Z 2 − 1 + ln(Z + Z 2 − 1)] . (10.59)
32πGρmax
In the vicinity of R + Rmin from (10.59) we find

4πG
, +
R(τ ) = Rmin 1+ ρmax · τ 2 . (10.60)
3
In the region Rmin << R < R0 we obtain
'1/4
32πG
&
R(τ ) = Rmin ρmax · τ 1/2 . (10.61)
3
In this region the dependence on time of the density of mat-
ter determined by equation (10.46), with account of (10.50),
(10.51) and (10.61), has the form
3
ρ(τ ) = , (10.62)
32πGτ 2
i.e. coincides with the known equation that yields the Fried-
man model in GRT for a “flat” Universe. We shall now de-
termine the time corresponding to transition from the stage of
expansion of the Universe dominated by radiation to the stage
dominated by nonrelativistic matter. According to (10.61) we
have '1/2
32πG
&
2 2
R0 = Rmin ρmax · τ0 . (10.63)
3
112
Hence, taking into account (10.44), (10.45) and (10.58) we find
5
ρ3/2 (τc ) 3
τ0 = r2 =
ρm (τc ) 32πG
5
3
= 2, 26 · 108 + 1, 5 · 1011 s . (10.64)
32πG
Now consider development of the Universe, when pressure
can be neglected. At this stage of evolution we write equation
(10.25) as
%2
2ωx2
$
dx
= (x − 1)[(α − x3 )(x2 + x + 1) − 3x2 ] . (10.65)
dτ α
6
Here x = Rmax /R, α = 2Rmax . Taking into account that
α >> 3 , (10.66)
we find
x0
α # dy
6
τ = τ0 + - . (10.67)
2ω x y (y 3 − 1)(x31 − y 3)
Here x0 = Rmax /R0 , x1 = 21/3 · Rmax

2
. Upon integration of
(10.67) we obtain
5
1 α
τ = τ0 + [arcsin f (x0 ) − arcsin f (x)]. (10.68)
3 2x1 ω
Here
(x31 + 1)x3 − 2x31
f (x) = . (10.69)
x3 (x31 − 1)
Note that
2
f (x0 ) + 1 −, (10.70)
x30
2 π 2
arcsin f (x0 ) + arccos 3/2 = − 3/2 . (10.71)
x0 2 x0
113
Taking into account (10.71) we find
2 1 π
, +
τ = τ0 − √ 3/2
+ √ − arcsin f (x) . (10.72)
3 2ωx0 3 2ω 2
Taking into account the equality

1
τ0 = √ 3/2
, (10.73)
2 2ωx0
expression (10.72) can be written in the form

√ π
3 2ω(τ + βτ0 ) = − arcsin f (x) . (10.74)
2
Hence, we have
(α + 1)x3 − 2α
cos λ(τ + βτ0 ) = . (10.75)
x3 (α − 1)
Here
5
√ mc2
$ %
3
λ = 3 2ω = , β = 1/3 . (10.76)
2 h̄
From expression (10.75) we find

, +1/6 ( )1/3
α (α + 1) − (α − 1) cos λ(τ + βτ0 )
R(τ ) = · . (10.77)
2 2α
Owing to the equality (10.40) the following relation occurs:

( )3
ρm (τ ) Rmax
= , (10.78)
ρmin R(τ )
taking into account (10.78) we obtain

2αρmin
ρm (τ ) = . (10.79)
(α + 1) − (α − 1) cos λ(τ + βτ0 )
114
Since α >> 1, from (10.79) we have
ρmin
ρm (τ ) = 2 λ(τ +βτ0 )
, (10.80)
sin 2
in a similar way from formula (10.78) we have
λ(τ + βτ0 )
R(τ ) = Rmax sin2/3 . (10.81)
2
λ(τ +βτ0 )
In the region of values 2
' 1 we have
1
ρm (τ ) = , (10.82)
6πG(τ + βτ0 )2
( )2/3
λ(τ + βτ0 )
R(τ ) = Rmax . (10.83)
2
For τ >> βτ0 formulae (10.82) and (10.83) yield for ρm (τ ) and
R(τ ) time dependencies similar to those obtained within the
Friedman model in GRT for a ”flat” Universe.
Making use of formulae (10.44), (10.45) and (10.51) one
can readily establish the following relation:
%1/12 %1/12
ρ1/3
$ $
m (τc ) ρmax −2 ρmax
Rmax = + 3, 6 · 10 . (10.84)
1/4
ρr (τc ) 4ρ2min ρ2min
In a similar manner with the aid of expression (10.57) and of

(10.50) one can express, via ρmax , the second turning point
Rmin :
$ %1/6
ρmin
Rmin = . (10.85)
2ρmax
From (10.84) and (10.85) it is clear that the existence of the
graviton mass not only removes the cosmological singularity,
but also stops the expansion process of the Universe, which
undergoes transition to the compression phase. Thus, the evo-
lution of a homogeneous and isotropic Universe is determined
115
by modern observational data (10.41), by the maximum den-
sity of matter and the graviton mass. The scalar curvature
is the largest at the beginning of the “expansion”, and on the
basis of (10.21), (10.24) and (10.85) equals the following value:
ρmax
Rµν g µν = −16πG · ,
c2
while its minimum value at the end of “expansion” is
'2
3 mc
&
µν
Rµν g = .
2 h̄
The initial acceleration, which serves as the initial “push”
that led to the “expanding” Universe, is, in accordance with
(10.24), (10.32) and (10.85), the following:
%1/3
d2 R 2
$
1 5/6 mc
= (8πGρmax ) .
dτ 2 3 2h̄
It arises at the moment when the density of matter stops grow-

ing during the preceding cycle. The maximum density of mat-
ter in the Universe in this model remains undefined. It is
related to the integral of motion. The latter is readily estab-
lished. We write equation (10.24) in the form
d2 R p 8πG 1
& ' & '
2
= −4πG ρ + 2 R+ ρR−2ω R − 5 . (10.86)
dτ c 3 R
Determining from equation
1 dR 1 dρ
=− p (10.87)
R dτ 3(ρ + c2
) dτ
7 8
p
the value of ρ + c2
we find
p 1 dρ
ρ+ 2
=− R . (10.88)
c 3 dR
116
Substituting this value into equation (10.86) we obtain
d2 R 4πG d d 1
& '
2
= · (ρR2 ) − ω R2 + . (10.89)
dτ 3 dR dR 2R4
Introducing the notation
4πG 2 1
& '
V =− ρR + ω R2 + , (10.90)
3 2R4
one can write equation (10.89) in the form of the Newton equa-
tion of motion
d2 R dV
= − , (10.91)
dτ 2 dR
where V plays the role of the potential. Multiplying (10.91)
by dR
dτ
we obtain
 $ %2 
d 1 dR
+V = 0. (10.92)
dτ 2 dτ
Hence, we have $ %2
1 dR
+V =E , (10.93)
2 dτ
where E is an intergal of motion, the analog of energy in clas-
sical mechanics. Comparing (10.93) with (10.25) and taking
into account (10.31) we obtain
%2
mc2
$
4 1
Rmax = . (10.94)
8E h̄
Substituting into (10.94) expression (10.84) we find

 7 810 1/3
mc2
h̄
E = 7, 4 · 104  . (10.95)
 
(16πG)2 ρmax

This quantity is extremely small.

117
Thus, ρmax is actually an integral of motion, determined by
the initial conditions of the dynamic system. The analysis per-
formed reveals that the model of a homogeneous and isotropic
Universe develops, in accordance with RTG, cyclically start-
ing from a certain finite maximum density ρmax down to the
minimum density, and so on. The Universe can only be “flat”.
Theory predicts the existence in the Universe of a large “hid-
den” mass of matter. The Universe is infinite and exists for
an indefinite time, during which an intense exchange of in-
formation took place between its regions, which resulted in
the Universe being homogeneous and isotropic, with a cer-
tain inhomogeneity structure. In the model of a homogeneous
and isotropic Universe this inhomogeneity is not taken into
account, for simplifying studies. The information obtained
is considered a zeroth approximation, that usually serves as a
background in considering the development of inhomogeneities
caused by gravitational instability. “Expansion” in a homo-
geneous and isotropic Universe, as we are convinced, is due
to variation of the gravitational field, and no motion of mat-
ter occurs, here. The existence of a certain inhomogeneity
structure in the distribution of matter in space introduces a
significant change, especially in the period after the recombi-
nation of hydrogen, when the Universe becomes transparent
and the pressure of radiation no longer hinders the collection
of matter in various parts of the Universe.
This circumstance results in the motion of matter relative
to the inertial reference system. Thus arise the peculiar velo-
cities of galaxies with respect to the inertial reference system.
A reference system related to the relic radiation can, with
a great precision, be considered inertial. Naturally, a refe-
rence system related to the relic gravitational radiation would
to an extremely high degree be close to an inertial system.
What was the maximum density of matter, ρmax , earlier in
the Universe? An attractive possibility is reflected in the hy-
pothesis that ρmax is determined by the world constants. In
this case, the Planck density is usually considered to be ρmax .
118
Here, however, there exists the problem of overproduction of
monopoles arising in Grand Unification theories. To overcome
this problem, one usually applies the “burning out” mech-
anism of monopoles during the inflational expansion process
due to the Higgs bosons. Our model provides another, alterna-
tive, possibility. The quantity ρmax may even be significantly
smaller, than the Planck density. In this case the tempera-
ture of the early Universe may turn out to be insufficient for
the production of monopoles, and the problem of their over-
production is removed in a trivial manner. This, naturally,
does not exclude the possibility of inflational expansion of the
Universe, if it turns out to be that at a certain stage of its
development the equation of state is p = −ρ.
Thus, in accordance with RTG, no pointlike Big Bang oc-
curred, and, consequently, no situation took place, when the
distance between galaxies were extremely small. Instead of the
explosion, at each point of the space there occurred a state of
matter of high density and temperature, and it further devel-
oped till the present moment, as described above. The differ-
ence between the development of a homogeneous and isotropic
Universe in RTG and GRT arose owing to the scaling factor
R(τ ) in RTG not turning into zero, while in GRT it becomes
zero at a certain moment in the past.
In GRT with the cosmological term λ the homogeneous and
isotropic model of the Universe is also possible in the absence
of matter. The solution of the GRT equations for this case
was found by de Sitter. This solution corresponds to curved
four-dimensional space-time. This signifies the existence of
gravity without matter. What is the source of this gravity?
Usually, it is vacuum energy identified with the cosmological
constant λ that is considered to be this source. In RTG, when
matter is absent (ρ = 0), in accordance with equations (10.25)
and (10.31), the right-hand side should turn to zero, which
is possible only if Rmax = 1. Hence it follows that R ≡ 1,
and, consequently, the geometry of space-time in the absence
of matter will be pseudo-Euclidean. Thus, in accordance with
119
RTG, when matter is absent in the Universe, there also exists
no gravitational field, and, consequently, vacuum possesses no
energy, as it should be. According to RTG, the Universe can-
not exist without matter.
In conclusion let us determine the horizon of particles and
the horizon of events. For a light ray, in accordance with the
interval (10.17), we have
dr 1
=√ . (10.96)
dτ aR(τ )
The distance covered by the light by the moment τ is
r(τ
# ) #τ
√ dσ
dr (τ ) = aR(τ ) dr = R(τ ) . (10.97)
R(σ)
0 0
If the gravitational field were absent, the distance covered by

the light would be cτ . As R we should have substituted ex-
pression (10.59) for the interval (0, τ0 ), and expression (10.81)
for the time interval (τ0 , τ ). We shall estimate dr (τ ) approxi-
mately by the expression
πτ
R(τ ) = Rmax sin2/3 (10.98)
2τmax
Throughout the entire integration interval
√
+2/3 #y
2τmax πτ dx
,
dr (τ ) = sin · √ =
π 2τmax x2/3 1 − x2
0
6τmax √ 1 1 7
& '
= yF , , ,y . (10.99)
π 2 6 6
Here y = sin2 2τπτ
max
, F (a, b, c, y) is a hypergeometric function.
We shall give the values for some quantities determining
the evolution of a homogeneous and isotropic Universe. We
set the graviton mass to m = 10−66 g, while the present-day
Hubble “constant” is
km
Hc + 74 . (10.100)
s Mpc
120
Then for the present-day moment of time τc , qc will be equal
to
g
τc + 3 · 1017 s, qc = 0, 59, ρc = 10−28 . (10.101)
cm3
According to formula (10.33) the half-period of cyclical devel-
opment is
τmax = 9π · 1017 s . (10.102)
It must be stressed that the parameters τc , qc determining
the evolution of the Universe are practically independent of the
maximum density of matter ρmax . The maximum temperature
(and, hence the maximum density), that could occur in the
Universe, may be determined by such phenomena, that took
place in these extreme conditions, and the consequences of
which may be observed today. A special role, here, is played
by the gravitational field, which contains the most complete
information on the extreme conditions in the Universe. In
the model, considered above, of an isotropic and homogeneous
Universe the known problems of singularities, of causality, of
flatness, that are present in GRT, do not arise.
Making use of (10.99) and (10.101) we find the size of the
observable part of the Universe at the moment τc :
dr (τc ) + 3cτc = 2, 7 · 1028 cm .
We see that the path covered by light in the gravitational field

of the Universe during the time τc is three times larger than
the corresponding distance in absence of the gravitational field,
cτc . During the half-period of evolution, τmax , the horizon of
particles be
cτmax Γ(1/6)
dr (τmax ) = √ · . (10.103)
π Γ(2/3)
The horizon of events is determined by the expression
#∞
dσ
dc = R(τ ) . (10.104)
τ
R(σ)
121
Since the integral (10.104) turns to infinity, the horizon of
events in our case does not exist. This means that information
on events taking place in any region of the Universe at the
moment of time τ will reach us. This information can be
obtained with the aid of gravitational waves, since they are
capable of passing through periods, when the density of matter
was high.
We shall especially note that within the framework of RTG
a homogeneous and isotropic Universe can exist only if the
graviton mass differs from zero. Indeed, in accordance with
(10.46) and (10.51), the constant A in expression (10.35) equals
7 82/3
ρmin
A = ρ1/3
max 2
. Therefore, if ρmax is fixed, then the con-
stant A turns to zero, when m = 0.
11. The gravitational field of a
spherically symmetric static
body
The issue of what takes place in the vicinity of the Schwar-

zschild sphere, when the graviton has a rest mass, was first
dealt with in relativistic theory of gravity in ref. [2], in which
the following conclusion was made: in vacuum the metric coef-
ficient of effective Riemannian space, g00 , on the Schwarzschild
sphere differs from zero, while g11 has a pole. These changes,
that in the theory are due to the graviton mass, result in the
“rebounding” effect of incident particles and of light from the
singularity on the Schwarzschild sphere, and consequently, in
the absence of “black holes”.
Further, in ref. [14] a detailed analysis of this problem in
RTG was performed, which clarified a number of issues, but
which at the same time revealed that the “rebounding” takes
place near the Schwarzschild sphere. In the present work we
follow the article [13], in which it was shown in a most simple
and clear manner that at the point in vacuum, where the met-
ric coefficient of effective Riemannian space g11 has a pole, the
other metric coefficient g00 does not turn to zero. The resul-
ting singularity cannot be removed by a choice of the reference
system, so the solution inside a body cannot be made to match
the external solution. In this case, if transition is performed
to the reference system related to a falling test body, then it
turns out to be that the test body will never reach the surface
of the body, that is the source of the gravitational field. Pre-
cisely this circumstance leads to the conclusion that the radius
of a body cannot be inferior to the Schwarzschild radius. All
this issue will be dealt with in detail in this section.
123
We now write equations (5.19), (5.20) in the form
1 1 mc 2 µ
& ' &
− δνµ R +
Rνµ δν + g µα γαν −
2 2 h̄
1 µ αβ
'
− δν g γαβ = κTνµ , (11.1)
2
Dµ g̃ µν = 0 . (11.2)
√
Here g̃ µν = −gg µν , g = det gµν , Rνµ is the Ricci tensor,
κ = 8πG
c2
, G is the gravitational constant, Dµ is the covariant
derivative in Minkowski space, γµν (x) is the metric tensor of
Minkowski space in arbitrary curvilinear coordinates.
Let us now determine the gravitational field created by a
spherically symmetric static source. The general form of an
interval of effective Riemannian space for such a source has
the form
ds2 = g00 dt2 + 2g01 dtdr + g11 dr 2 + g22 dΘ2 + g33 dΦ2 , (11.3)
We introduce the notation
g00 (r) = U(r), g01 (r) = B(r),

B 2 (r)
( )
g11 (r) = − V (r) − , (11.4)
U(r)
g22 (r) = −W 2 (r), g33 (r, Θ) = −W 2 (r) sin2 Θ.
The components of the contravariant metric tensor are
B2
$ %
00 1 B 1
g (r) = 1− , g 01 (r) = −
, g 11 (r) = − ,
U UV UV V
(11.5)
1 1
g 22 (r) = − 2 , g 33 (r, Θ) = − 2 2 .
W W sin Θ
The determinant of the metric tensor gµν is
g = detgµν = −UV W 4 sin2 Θ . (11.6)

124
For a solution to have physical meaning the following condition
must be satisfied:
g<0. (11.7)
In the case of spherical coordinates g can turn to zero only
at the point r = 0. On the basis of (11.5) and (11.6) we find
the density components of the metric tensor
√
g̃ µν = −gg µν . (11.8)
One have the form

W2 B2 BW 2
$ %
g̃ 00 =√ V − sin Θ, g̃ 01 = − √ sin Θ,
UV U UV
5
U 2
g̃ 11 = − W sin Θ, (11.9)
V
√
22
√ 22
√ 33 UV
g̃ = − UV sin Θ, g̃ = − UV sin Θ, g̃ = − .
sin Θ
We shall carry out all reasoning in an inertial reference
system in spherical coordinates. An interval in Minkowski
space has the form
dσ 2 = dt2 − dr 2 − r 2 (dΘ2 + sin2 ΘdΦ2 ) . (11.10)
The Christoffel symbols in Minkowski space that differ from

zero and that are determined by the formula
λ 1
γµν = γ λσ (∂µ γσν + ∂ν γσµ − ∂σ γµν ) (11.11)
2
are equal to
1 1 1
γ22 = −r γ33 = −r sin2 Θ, γ12
2 3
= γ13 = ,
r (11.12)
2 3
γ33 = − sin Θ cos Θ, γ23 = cot Θ .
We write equation (11.2) in an expanded form,
Dµ g̃ µν = ∂µ g̃ µν + γλσ
ν λσ
g̃ = 0 . (11.13)
125
In Galilean coordinates of Minkowski space they have the form
∂µ g̃ µν = 0 . (11.14)
In the case of a static gravitational field we have from (11.14)
∂i g̃ iν = 0, i = 1, 2, 3 . (11.15)
Applying the tensor transformation law, it is possible to

express the components g̃ i0 in Cartesian coordinates in terms
of the components in spherical coordinates
i0 BW 2 xi √ √
g̃ = − √ · 3 , −g = UV W 2 r −2 . (11.16)
UV r
Here xi are spatial Cartesian coordinates. Assuming ν = 0
in (11.15) and integrating over the spherical volume upon ap-
plication of the Gauss-Ostrogradsky theorem, we obtain the
following integral over the spherical surface:
BW 2
? ?
g̃ i0 dsi = − √ (0xd0s) = 0 . (11.17)
r 3 UV
Taking into account the equality
?
(0xd0s) = 4πr 3 , (11.18)
we obtain
BW 2
√ =0. (11.19)
UV
Since equation (11.14) holds valid both within matter and
outside it, (11.19) should be valid for any value of r. But
since, owing to (11.7), U, V and W cannot be equal to zero,
then from (11.19) it follows that
B=0. (11.20)
126
The interval (11.3) of effective Riemannian space assumes the
form
ds2 = Udt2 − V dr 2 − W 2 (dΘ2 + sin2 ΘdΦ2 ) . (11.21)
From (11.20) it follows that no static solution exists of

the Hilbert–Einstein equations in harmonic coordinates, that
contains in the interval a term of the form
B(r)dtdr . (11.22)
The energy-momentum tensor of matter has the form

p p
& '
Tνµ = ρ + 2 v µ vν − δνµ · 2 . (11.23)
c c
In expression (11.23) ρ is the mass density of matter, p is the
isotropic pressure, and
dxµ
vµ = (11.24)
ds
is the four-velocity satisfying the condition
gµν v µ v ν = 1 . (11.25)
From equations (11.1) and (11.2) follows
∇µ Tνµ = 0 , (11.26)
where ∇µ is the covariant derivative in effective Riemannian

space with the metric tensor gµν . In the case of a static body
1
v i = 0, i = 1, 2, 3; v 0 = √ , (11.27)
U
and therefore
p(r)
T00 = ρ(r), T11 = T22 = T33 = − ,
c2
(11.28)
Tνµ = 0, µ -= ν .
127
For the interval (11.21) the Christoffel symbols, differing
from zero, are
1 dU 1 dU 1 dV
Γ001 = , Γ100 = , Γ111 = ,
2U dr 2V dr 2V dr
W dW 1
Γ122 =− , Γ = sin2 Θ · Γ122 , (11.29)
V dr 33
1 dW
Γ212 = Γ313 = , Γ233 = − sin Θ cos Θ , Γ323 = cot Θ.
W dr
Applying the following expression for the Ricci tensor:
Rµν = ∂σ Γσµν − ∂ν Γσµσ +
+Γσµν Γλσλ − Γσµλ Γλσν , Rνµ = g µλ Rλν (11.30)
and substituting into it expressions for the Christoffel sym-

bols from (11.29), it is possible to reduce equations (11.1) for
functions U, V and W to the form
%2
2 d2 W
$
1 1 dW 1 dW d 1
& '
2
− 2
− 2
− +
W & V' W ( dr & V W' dr 2 ) W dr dr V
1 mc 2 1 1 1 r
+ 1+ − − 2 = κρ , (11.31)
2 h̄ 2 U V W
$ %2
1 1 dW 1 dW dU
− − +
W 2 V W 2 dr UV W dr dr
r2
' ( )
1 mc 2 1 1 1 p
& & '
+ 1− − − 2 = −κ 2 , (11.32)
2 h̄ 2 U V W c
1 1 1 1
− W !! − U !! + W ! !
V + (U ! )2 +
VW 2UV 2W V 2 4V U 2
1 ! ! 1
+ U V − W !U ! +
4UV 2 2UV W
'2 ,
1 mc 1 1 1 p
& & '+
+ 1− + = −κ . (11.33)
2 h̄ 2 U V c2
128
Equation (11.13), with account of (11.12), (11.9) and (11.20),
can be reduced to the form
5 
d  U 2 √
W = 2r UV . (11.34)
dr V
We note that by virtue of the Bianchi identity and of equa-
tion (11.2) one of the equations (11.31) – (11.33) is a conse-
quence of the other ones. We shall further take equations
(11.31), (11.32) and (11.34) to be independent.
We write equation (11.26) in an expanded form
∇µ Tνµ ≡ ∂µ Tνµ + Γµαµ Tνα − Γαµν Tαµ = 0 . (11.35)
Making use of expressions (11.28) and (11.29) we obtain
1 dp ρ + cp2 dU
· =− · . (11.36)
c2 dr 2U dr
Taking into account the identity
 $ %2  $ %2
1 d W dW 1 dW
7 8·  = +
W 2 dW dr V dr V W2 dr
dr
2 d2 W 1 dW d 1
& '
+ 2
+ , (11.37)
V W dr W dr dr V
equation (11.31) can be written as
 
'2 (
d  W 1 mc
&
1− 82  + W 2 − r2 +

dW V dr 2 h̄
 7
dW
W2
')
1 1
&
+ − = κW 2 ρ . (11.38)
2 U V
In a similar manner we transform equation (11.32):
'2 ,
W d 1 mc
&
1− 82 ln(UW ) + W 2 − r2−
dW 2 h̄
7
dr
V dW
1 1 1 W 2p
& '+
− − = −κ . (11.39)
2 U V c2
129
We write equations (11.34) and (11.36) in the form
 5 
d  2 U √ dr
W = 2r UV . (11.40)
dW V dW
1 dp p 1 dU
& '
· = − ρ + · . (11.41)
c2 dW c2 2U dW
In equations (11.38) – (11.41) we pass to dimensionless
variables. Let l be the Schwarzschild radius of the source, the
mass of which equals M, then
2GM
l= . (11.42)
c2
We introduce new variables x and z, equal to
W = lx, r = lz. (11.43)
Equations (11.38) – (11.41) assume the form

 
d  x 
,
2 2
1−  7 82  + % x − z +
dx V dz
dx
1 1 1
& '+
+ x2 − = κ̃x2 ρ(x), (11.38a)
2 U V
(
d x
1 − 7 82 ln(xU) + % x2 − z 2 −
dz
V dx dx
x2 x2 p(x)
')
1 1
&
− − = −κ̃ , (11.39a)
2 U V c2
 5 
d  2 U dz √
x = 2z UV , (11.40a)
dx V dx
1 dp p 1 dU
& '
2
=− ρ+ 2 . (11.41a)
c dx c 2U dx
130
Here % is a dimensionless constant equal to
'2
1 2GMm
&
%= , κ̃ = κl2 . (11.44)
2 h̄c
The sum of and the difference between equations (11.38a)
and (11.39a) are
 
d  x  x d
2 −  7 82  − 7 82 ln(xU) +
dx V dz dz
V dx dx
dx
p
& '
2 2 2
+ 2%(x − z ) = κ̃x ρ − 2 , (11.45)
c
 
d  x  x d
 7 82  − 7 82 ln(xU) −
dx V dz V dz dx
dx dx
1 1 p
& ' & '
2 2
− %x − = −κ̃x ρ + 2 . (11.46)
U V c
We introduce the new functions A and η:
1 x
U= , V = 7 82 . (11.47)
xηA dz
A dx
In these new variables equation (11.45) assumes the form
d ln η p
& '
A + 2 + 2%(x2 − z 2 ) = κ̃x2 ρ − 2 . (11.48)
dx c
Equation (11.38a) is written in the form
dA x2 1 1
& '
= 1 + %(x2 − z 2 ) + % − − κ̃ · x2 ρ(x) . (11.49)
dx 2 U V
In accordance with the causality condition (see Addendum)
γµν U µ U ν = 0 , (11.50)
131
gµν U µ U ν ≤ 0 , (11.50a)
it is easy to establish the inequality
U ≤V . (11.51)
For our problem it suffices to consider only the values of x and

z from the interval
1 1
0≤x' √ , 0≤z' √ . (11.52)
2% 2%
These inequalities impose an upper limit on r, W :
h̄
r, W ' . (11.53)
mc
In the case of such a restriction equation (11.49) assumes the
form
dA x2 1 1
& '
=1+% − − κ̃x2 ρ(x) . (11.54)
dx 2 U V
Outside matter we have
dA x2 1 1
& '
= 1+% − . (11.55)
dx 2 U V
By virtue of causality (11.51) the following inequality holds
valid beyond matter
dA
≥ 1. (11.56)
dx
Integrating (11.54) over the interval (0, x) we obtain
#x #x
% 1 1
& '
!2 ! 2
A(x) = x + x − dx − κ̃ x! ρ(x! )dx! . (11.57)
2 U V
0 0
A(0) in (11.57) is set equal to zero, since if it were different

from zero, function V (x) would turn to zero as x tends toward
zero, which is inadmissible from a physical standpoint. On
132
the basis of (11.56) function A(x) beyond matter grows in
monotonic way with x, and it therefore can only have the sole
root
A(x1 ) = 0, x1 > x0 . (11.58)
On the basis of (11.57) we have
#x1
% 1 1
& '
2
x1 = 1 − x! − dx! . (11.59)
2 U V
0
We have here taken into account that when l is chosen to be

equal to (11.42)
#x0
2
κ̃ x! ρ(x! )dx! = 1.
0
The matter is concentrated inside the sphere 0 ≤ x ≤ x0 .

We shall further consider the case, when the radius of the
body, x0 , is less than x1 . Precisely in this case in vacuum, i.e.
outside the body, there will exist a singularity which cannot
be removed by a choice of reference system.
Owing to the graviton mass the zero of function A is shifted
inward the Schwarzschild sphere. Since as x tends toward
x1 , V (x) tends toward infinity, owing to A(x) tending toward
zero, there will exist such a vicinity about the point x1
x1 (1 − λ1 ) ≤ x ≤ x1 (1 + λ2 ), λ1 > 0, λ2 > 0 , (11.60)
(λ1 and λ2 assume small fixed values), inside which the follow-
ing inequality holds valid:
1 1
( . (11.61)
U V
In this approximation we obtain
#x
% 2 1
A(x) = x − x1 + dx! x! . (11.62)
2x U
1
133
Substituting into this expression U in the form (11.47) we find
#x
% 3
A(x) = x − x1 + dx! x! η(x! )A(x! ). (11.63)
2x
1
In the region of variation of x one can substitute x31 for x3

within the interval (11.60):
#x
%
A(x) = x − x1 + x31 η(x! )A(x! )dx! . (11.64)
2 x
1
Hence, we obtain
dA %
= 1 + x31 η(x)A(x). (11.65)
dx 2
In the considered approximation (11.52) equation (11.48) as-
sumes the form
d ln η
A +2=0. (11.66)
dx
We now introduce a new function
x31
f (x) = η(x)A(x) . (11.67)
2
Equation (11.65) assumes the form
dA
= 1 + %f (x) , (11.68)
dx
and equation (11.66) assumes the form
A df dA
· − = −2 . (11.69)
f dx dx
From equations (11.68) and (11.69) we find
(1 − %f )f
A(x) = − 7
df
8 . (11.70)
dx
134
From expression (11.67) we obtain
df
2 dx
η(x) = − . (11.71)
x31 (1 − %f )
Substituting (11.70) and (11.71) into (11.47) we find

df
x31 x dx
U= , V =− 7 82 . (11.72)
2xf f (1 − %f ) dz dx
Making use of these expressions we can rewrite the determi-

nant g as
df 4
x31 dx x
g= 7 82 sin2 Θ < 0 . (11.73)
dz
2f 2 dx
(1 − %f )
For condition (11.7) to be satisfied, it is necessary for ex-

df
pressions dx and (1 − %f ) to have opposite signs. Substituting
(11.70) into (11.68) we obtain
B B
d B df B d 1 + %f df
ln B B − ln |f (1 − %f )| = · . (11.74)
B B
dx B dx B dx f (1 − %f ) dx
Hence, we find
B (1 − %f ) df B
B B
d dx B
ln BB 2
B=0. (11.75)
dx B f B
Thus,
B (1 − %f ) df B
B B
dx B
B
2
= C0 > 0 . (11.76)
f
B B
B B
df
Taking into account that the quantities (1 − %f ) and dx
must
have opposite signs we find
df C0 f 2
=− . (11.77)
dx (1 − %f )
135
(1 − %f )2 1
A(x) = , A(x1 ) = 0 at f = . (11.78)
C0 f %
With account of (11.78), expression (11.47) for the function V

assumes the form
C0 xf
V = 7 82 . (11.79)
dz
(1 − %f )2 dx
Integrating (11.77) and taking into account (11.78) we obtain

1
C0 · (x − x1 ) = + % ln %|f | − % . (11.80)
f
Relation (11.80) has been obtained for the domain of x

values determined by equalities (11.60), however, it is also
valid in the region where the graviton mass can be neglected.
In accordance with (11.60) the domain C0 (x−x1 ) is within
the limits
− C0 x1 λ1 ≤ C0 (x − x1 ) ≤ C0 x1 λ2 , (11.81)
when f is positive, it satisfies the inequalities

1
C̃ ≤ f ≤ . (11.82)
%
Making use of (11.80), in accordance with (11.81), we have
1
+ % ln %f − % ≤ C0 x1 λ2 .
f
Hence, we can find C̃:

1
+ % ln %C̃ − % = C0 x1 λ2 . (11.83)
C̃
136
From expression (11.83) we find the approximate value for C̃:
1
C̃ = . (11.84)
C0 x1 λ2
For negative values of f , the value |f |, determined from
the following equation, corresponds to the point x = x1 :
1
− + % ln %|f | − % = 0 . (11.85)
|f |
Hence, we find
a 1+a
|f | = , ln a = . (11.86)
% a
In accordance with (11.81), the following inequality should be
satisfied:
1
− C0 x1 λ1 ≤ − + % ln %|f | − % . (11.87)
|f |
Hence it is possible to find the lower boundary for |f | = D:
1
− C0 x1 λ1 = − + % ln %D − % . (11.88)
D
From expression (11.88) we find the approximate value for D:
1
D= . (11.89)
C0 x1 λ1
This means that the quantity |f | satisfies the inequality
1
|f | ≥ D = . (11.89a)
C0 x1 λ1
Let us now establish the dependence of variable z upon

x. Substituting (11.47) into (11.40a) and taking into account
(11.48) we obtain
$ %
d dz dz *
A x = 2z − x 1 + %(x2 − z 2 )−
dx dx dx
1 2 p
& '+
− κ̃x ρ − 2 . (11.90)
2 c
137
In the approximation (11.52), outside matter, equation (11.90)
assumes the form
$ %
d dz dz
A x + x − 2z = 0 . (11.91)
dx dx dx
We have to find the regular solution z(x) of equation (11.91).

In equation (11.91) we pass from variable x to f . Applying
relation (11.80)
1
x= [C0 x1 f + 1 − %f + %f ln %|f |] , (11.92)
C0 f
and taking into account (11.65), (11.66) and (11.83), equation
(11.91) can be represented in the form
d2 z C0 xf + %f − 1 dz 2z
2
+ 2
· − =0. (11.93)
df C0 f x df C0 f 3 x
By direct substitution one can establish that the expression
x1 1
z= + [1 − %f + %f ln %|f |] (11.94)
2 C0 f
satisfies equation (11.93) with an accuracy up to the quantity
(1 − %f + ln %|f |)
% , (11.95)
C02 xf 3
that is extremely small in the vicinity of the point x1 . From
expressions (11.92) and (11.94) we find
x1
z =x− . (11.96)
2
Taking this relation, as well as (11.79) and (11.72), into ac-
count we obtain
x31 C0 xf
U= , V = . (11.97)
2xf (1 − %f )2
138
For negative values of f the causality condition (11.51) as-
sumes the form
|f |2 (2x2 C0 − %2 x31 ) − 2%x31 |f | − x31 ≤ 0. (11.98)
Inequality (11.98) is not satisfied, since it does not com-

ply with inequality (11.89a). Thus, the causality principle is
violated in the domain of negative values of f . This means
that in the region x1 (1 − λ1 ) ≤ x < x1 the solution has no
physical sense. If x0 < x1 (1 − λ1 ), the situation arises, when
the physical solution inside the body, 0 ≤ x ≤ x0 , cannot be
made to match the physical solution in the region x > x1 ,
since there exists an intermediate region x1 (1 − λ1 ) ≤ x < x1 ,
within which the solution does not satisfy the causality princi-
ple. Hence it necessarily follows that x0 ≥ x1 . From a physical
point of view, it is necessary to exclude the equality x0 = x1 ,
since the solution inside the body should undergo smooth tran-
sition to the external solution. Consequently, the variable
f only assumes positive values. For values from the region
x ≥ x1 (1 + λ2 ) one may, in equations (11.38a) and (11.39a),
drop the terms containing the small parameter %. Thus, we
arrive at Schwarzschild’s external solution
( )
b x − 2ω
zs = (x − ω) 1 + ln , (11.99)
2ω x
x x − 2ω
Vs = 7 82 , Us = . (11.100)
dz
(x − 2ω) x
dx
Here, "ω" and "b" are certain constants that are determined
from the condition that solution (11.96), (11.97) is made to
match solution (11.99), (11.100). At point x = x1 (1 + λ2 ) the
function z from (11.96) is
1
& '
z = x1 + λ2 , (11.101)
2
139
At the same point zs equals
( )
b x1 (1 + λ2 ) − 2ω
zs = [x1 (1 + λ2 ) − ω] 1 + ln . (11.102)
2ω x1 (1 + λ2 )
From the condition that (11.101) and (11.102) match we find
x1
ω= , b=0. (11.103)
2
At point x = x1 (1 + λ2 ) the function U from (11.97) equals
x31
U= , (11.104)
2x1 (1 + λ2 )C̃
since C̃, in accordance with (11.84), is

1
C̃ = . (11.105)
C0 x1 λ2
C0 x31 λ2
U= , (11.106)
2(1 + λ2 )
At the same point, with account of (11.103), Us is
λ2
Us = . (11.107)
1 + λ2
From the condition that (11.106) and (11.107) match we
find
2
C0 = 3 . (11.108)
x1
At point x = x1 (1 + λ2) the function V from (11.97) equals
V = C0 x1 (1 + λ1 )C̃ . (11.109)
Substituting into (11.109) the value C̃ from (11.105) we obtain

1 + λ2
V = (11.110)
λ2
140
at the same point, with account of (11.99) and (11.103), Vs
equals
1 + λ2
Vs = , (11.111)
λ2
i.e. the solution for V matches the solution for Vs .
Thus, if the radius of a body exceeds the Schwarzschild ra-
dius, then the graviton mass can be neglected, and the interval
of effective Riemannian space in an inertial reference system
in spherical coordinates outside the body in the region (11.53)
has the form:
r − GM 2 r + GM 2
ds2 = dt − dr −
r + GM r − GM
−(r + GM)2 [(dΘ)2 + sin2 Θ(dϕ)2 ].
This expression is determined unambiguously from the com-

plete set of equations (11.1) and (11.2), and, here, there exists
no arbitrariness. When the solution inside the body is made
to match the solution outside the body it is also necessary, as
first shown by R. Avakian, to take into account the logarith-
mic term (11.99) which arises when the solution of equations
(11.2) is sought. However, since the radius of the Sun exceeds
the Schwarzschild radius significantly, we can do not take it
into account in calculations of gravitational effects in the Solar
system.
Now consider (11.92) for values of %f close to unity:
1 y
f= 7 y
8, ' 1. (11.112)
% 1+ - %
Substituting this expression into (11.92) and expanding in y-

we obtain
y 2 = 2%C0 (x − x1 ). (11.113)
141
Inequality (11.112) signifies, that the quantity (x−x1 ) = δ ' %,
i.e. 5
y - x − x1
= 2C0 · '1. (11.114)
% %
Substituting (11.113) into (11.112), and then f into (11.97),
we obtain for U and V the following expressions:
-
x31 [% + 2%C0 (x − x1 )]
U= ,
2x
- (11.115)
x[% + 2%C0 (x − x1 )]
V = .
2%(x − x1 )
Hence, within the domain of variable x satisfying inequality
(11.114), we have
%x31 x
U= , V = . (11.116)
2x 2(x − x1 )
We see that the presence of the graviton mass essentially
alters the character of the solution in the region close to the
gravitational radius. At the point, where the function V , in
accordance with (11.116), has a pole, the function U differs
from zero, while in general relativity theory it equals zero. It
is precisely owing to this circumstance, that an irreversible
gravitational collapse arises in GRT, during which there ap-
pear “black holes” (objects that have no material boundaries
and that are “cut off” from the external world). In RTG “black
holes” are impossible.
If one takes into account (11.42), (11.43), (11.96) and ne-
glects the second term in (11.59), then expressions (11.116)
for U and V assume the form:
'2 GM
GMm 1 r+
&
c2
U= , V = · GM , (11.117)
h̄c 2 r− c2
142
which coincides with the formulae of [2]. We note that the
residue at the pole of function V at % -= 0 equals GM c2
, while
2GM
at % = 0 it equals c2 . This is so, because, when % = 0, the
pole of function V at point x = x1 is due to function f , which
at this point has a pole, while, if % -= 0, it is due to function
(1 − %f ), which, in accordance with (11.92), turns to zero at
the point x = x1 .
We shall now compare the character of motion of test bod-
ies in effective Riemannian space with the metric (11.117) and
with the Schwarzschild metric. We write the interval (11.21)
of Riemannian space in the form
ds2 = Udt2 − Ṽ dW 2 − W 2 (dΘ2 + sin2 ΘdΦ2 ) . (11.118)
Here Ṽ is $ %2
dr
Ṽ (W ) = V . (11.119)
dW
The motion of a test body proceeds along a geodesic line of
Riemannian space
dv µ
+ Γµαβ v α v β = 0 , (11.120)
ds
where
dxµ
vµ = , (11.121)
ds
the velocity four-vector v µ satisfies the condition
gµν v µ v ν = 1 . (11.122)
Now consider radial motion, when
vΘ = vΦ = 0 . (11.123)
Taking into account (11.29), from equation (11.120) we find

dv 0 1 dU 0 1
+ · v v =0, (11.124)
ds U dW
143
where
dW
v1 = . (11.125)
ds
From equation (11.124) we find
d
ln(v 0 U) = 0 . (11.126)
dW
Hence, we have
dx0 U0
v0 = = , (11.127)
ds U
where U0 is the integration constant.
Taking into account (11.127), condition (11.122) for radial
motion assumes the form
%2
U02
$
dW
− 1 = Ṽ · . (11.128)
U ds
If we assume the velocity of a falling test body to be zero at

infinity, then we obtain U0 = 1. From (11.128) we find
5
dW 1−U
=− . (11.129)
ds U Ṽ
Taking into account (11.79), (11.96), (11.97) and (11.108) we
have
x3 2xf
U = 1 , Ṽ = 3 .
2xf x1 (1 − %f )2
Substituting these expressions into (11.129) we obtain
dW √
= − 1 − U (1 − %f ) . (11.130)
ds
Applying (11.108), (11.112) and (11.113), in the vicinity of the
point x1 we have
5
dW 2 x − x1
=− . (11.131)
ds x1 %x1
144
Passing from the variable x to W , in accordance with (11.43)
and taking into account (11.44), we obtain
5
dW h̄c2 W 2GM
& '
=− 1− 2 . (11.132)
ds mGM GM cW
Hence there evidently arises a turning point. Differentia-
ting (11.132) with respect to s we find
%2
d2 W h̄c2
$
1
= . (11.133)
ds2 2GM mGM
At the turning point, the acceleration (11.133) is very large,

and it is positive, i.e. there occurs repulsion. Integrating
(11.132), we obtain
%2
h̄c2
$
2GM 1
W = + · (s − s0 )2 . (11.134)
c2 2mGM GM
Formulae (11.132) – (11.134) coincide with the formulae

of ref. [2]. The presence of the Planck constant in formula
(11.132) is due to the wave nature of matter, in our case,
of gravitons exhibiting rest mass. From formula (11.134) it
is evident that a test body cannot cross the Schwarzschild
sphere. In GRT the situation is totally different. From the
Schwarzschild solution and expression (11.129) it follows that
a test body will cross the Schwarzschild sphere, and that a
“black hole” will form. Test bodies or light can only cross
the Schwarzschild sphere inwards, and then they can never
leave the Schwarzschild sphere any more. The same result is
obtained, if we pass to a synchronous set of freely falling test
bodies with the aid of the transformations
( )1/2
Ṽ (1 − U)
#
τ =t+ dW . (11.135)
U
145
( )1/2
#
Ṽ
R =t+ dW . (11.136)
U(1 − U)
In this case the interval (11.118) assumes the form
ds2 = dτ 2 − (1 − U)dR2 − W 2 (dΘ2 + sin2 ΘdΦ2 ) . (11.137)
In this form, the singularities of the metric coefficients dis-
appear both for the Schwarzschild solution, when % = 0, and
for the solution in our case, when % -= 0. However, while in
GRT the variable W may turn to zero, in RTG, by virtue of
expression (11.134), it is always larger, than the Schwarzschild
radius.
Subtracting from expression (11.136) expression (11.135)
we obtain 2
3
# 3 U Ṽ
R − τ = dW 4 . (11.138)
(1 − U)
Differentiating equality (11.138) with respect to τ , we find
5
dW (1 − U)
=− . (11.139)
dτ U Ṽ
Thus, we arrive at the same initial equation (11.129). Taking
into account that r = W − GM c2
, on the basis of expressions
(11.117), we obtain from equation (11.139) the following:
%2
h̄c2 (R − cτ )2
$
2GM 1
W = 2
+ · (11.134a)
c 4 GMm GM
Hence, it is also evident that, if % -= 0, then a falling test body
can never cross the Schwarzschild sphere. In that case, when
% = 0, the Schwarzschild singularity in the metric does not
influence the motion of the test body in a falling synchronous
reference system. In GRT the following expression will occur,
instead of formula (11.134a):
+2/3 & '1/3
3 2GM
,
W = (R − cτ ) ,
2 c2
146
which testifies that a test body will reach the point W = 0 in
a finite interval of proper time. The falling particles, here, will
only cross the Schwarzschild sphere in one direction, inward.
We shall now calculate the propagation time of a light signal
from a certain point W0 to the point W1 = 2GM c2
, given by the
clock of a distant observer. From the expression ds2 = 0 we
have the following for the Schwarzschild solution:
dW 2GM
& '
= −c 1 − 2 . (11.140)
dt cW
Integrating this equation we obtain
2GM W0 − 2GM c2
W0 − W + ln = c(t − t0 ) . (11.141)
c2 W − 2GM
c2
Hence, it is obvious that an infinite time, by the clock of a

distant observer, is required in GRT in order to reach the
gravitational radius W1 = 2GM c2
. In RTG, as we established
earlier, the Schwarzschild solution is valid up to the point W =
W1 (1 + λ2 ), so the time required to reach this point is
2GM W0 − 2GM c2
c(t − t0 ) = W0 − W1 (1 + λ2 ) + ln . (11.142)
c2 λ2 2GM
c2
The propagation time of a light ray from the point W =

W1 (1 + λ2 ) to the point W1 can be calculated making use of
formulae (11.97) and (11.108). Within this interval we have
dW x3
= −c 1 (1 − %f ). (11.143)
dt 2xf
Hence, upon integration and a change of variable, we obtain
#1/-
2MG xdf
= c(t1 − t) . (11.144)
c2 f
f
147
In accordance with (11.84) and (11.108) the lower integration
limit is
x2
f = C̃ = 1 . (11.145)
2λ2
The integral (11.144) is readily calculated and with a good
accuracy leads to the following relation:
2GM 2λ2
c(t1 − t) = W1 λ2 + ln . (11.146)
c2 %
On the basis of (11.142) and (11.146), the time required
for a light signal to cover the distance between the points W0
and W1 = 2GM c2
, is equal to the sum of expressions (11.142)
and (11.146),
2GM W0 − 2GM c2
c(t1 − t0 ) = W0 − W1 + ln . (11.147)
c2 % GM
c 2
Hence it is seen, that in RTG, unlike GRT, the propagation

time of a light signal to the Schwarzschild sphere is finite, even
if measured by the clock of a distant observer. From formula
(11.147) it is evident that the propagation time is not enhanced
significantly by the influence of the gravitational field.
On the basis of the above presentation it is clear that, if
the graviton mass exists, % -= 0, then the solution in RTG
differs essentially from the Schwarzschild solution owing to
the presence on the Schwarzschild sphere of a singularity, that
cannot be removed by a choice of reference system. Thus, in
the case we have considered, when the radius of a body is
smaller than the Schwarzschild radius, or to be more precise,
when x0 < x1 , a test particle can never reach the surface of
the body, by virtue of (11.134). Owing to the presence of a
singularity, the physical condition g < 0 is violated outside
the body, and precisely for this reason a physical solution for
a static spherically symmetric body is possible only in the
148
case, when the point x1 is inside the body. This conclusion is
conserved, also, for a synchronous reference system, when the
metric coefficients (see (11.134a)) are functions of time.
Thus, in accordance with RTG, no Schwarzschild singula-
rity for a body of arbitrary mass exists, since the radius of the
body is greater than the Schwarzschild radius, and so the for-
mation of “black holes” (objects without material boundaries
and “cut off” from the external world) is impossible. This
conclusion complies with the conclusion made by A. Einstein
in 1939, most likely based on his physical intuition, than on
GRT logic. He wrote: “Schwarzschild’s singularity does not
exist, since matter cannot be concentrated in an arbitrary man-
ner; otherwise clustering particles would achieve the velocity of
light” 31 . A. Einstein, naturally, saw that the existence of the
Schwarzschild singularity violated his main principle: “to ac-
knowledge all conceivable (we shall not, here, deal with certain
restrictions, following from the requirement of uniqueness and
continuity) reference systems to be essentially equivalent for
describing nature” 32 . Precisely for this reason, he considered,
from a physical point of view, that no Schwarzschild singular-
ity in the metric coefficients should exist in a reference system
related to a distant observer, also. All this, however, is realized
in RTG, but not in GRT.
In accordance with RTG, as a field theory of gravity, a body
of arbitrary mass cannot undergo compression indefinitely, and
therefore no gravitational collapse involving the formation of a
“black hole” is possible. This means that a collapsing star
cannot go beneath its gravitational radius. Spherically
symmetric accretion of matter onto such a body, at its final
31
Einstein A. Collection of scientific works, M.: Nauka, 1966, vol.2,
art.119, p.531.
32
art.38, p.459.
149
stage of evolution (when nuclear resources are exhausted), will
be accompanied by a great release of energy, owing to mat-
ter falling onto the surface of the body. According to RTG,
gravitational absorption of light is impossible. In GRT, when
spherically symmetric accretion of matter onto a “black hole”
takes place, the energy release is quite low, since the falling
matter brings energy into the “black hole”. Gravitational ab-
sorption of light takes place. Gravitational self-closure of the
object occurs. Observational data on such objects could pro-
vide the answer, as to what happens with stars of large mass
at their final stage of evolution, when all nuclear resources are
exhausted.
Addendum
In spherical coordinates of Minkowski space the intervals of

Minkowski space and of effective Riemannian space have the
form
dσ 2 = dt2 − dr 2 − r 2 (dΘ2 + sin2 ΘdΦ2 ) , (1)
ds2 = U(r)dt2 − V (r)dr 2 − W 2 (r)(dΘ2 + sin2 ΘdΦ2 ) . (2)
We now introduce the velocity vector
dxi
vi = , v i = vei , (xi = r, Θ, Φ) . (3)
dt
ei represents the unit vector with respect to the metric of the
spatial part of Minkowski space
κik ei ek = 1 . (4)
In the general case κik is

γ0i γ0k
κik = −γik + . (5)
γ00
In case (1)
κik = −γik . (6)
Condition (4) for metric (1) has the form
(e1 )2 + r 2 [(e2 )2 + sin2 Θ · (e3 )2 ] = 1 . (7)
We define the velocity four-vector by the equality
v µ = (1, vei ) (8)
and require that it be isotropic in Minkowski space
γµν v µ v ν = 0 . (9)
151
Substituting (8) into (9) and taking into account (7) we find
v=1. (10)
Thus, the isotropic four-vector v µ is equal to
v µ = (1, ei ) . (11)
Since, in accordance with special relativity theory, motion

always proceeds inside or on the boundary of the Minkowski
causality cone, then, in the case of a gravitational field, the
causality principle
gµν v µ v ν ≤ 0 (12)
will be valid, i.e.
U − V (e1 )2 − W 2 [(e2 )2 + (e3 )2 sin2 Θ] ≤ 0 . (13)
Taking into account (7), expression (13) may be written as
W2 W2
$ %
U − 2 − V − 2 (e1 )2 ≤ 0 . (14)
r r
Let
W2
V − ≥0. (15)
r2
Owing to arbitrariness, 0 ≤ (e1 )2 ≤ 1, inequality (14) will be
satisfied only if
W2
U− 2 ≤0. (16)
r
From inequalities (15) and (16) follows
U ≤V . (17)
In the case, when

W2
V − <0, (18)
r2
152
we write inequality (14) as
W2
$ %
U −V − −V (1 − (e1 )2 ) ≤ 0 . (19)
r2
Owing to the arbitrariness of e1 , expression (19) will hold valid

for any values 0 ≤ (e1 )2 ≤ 1, only if
U ≤V . (20)
Thus, the causality principle in RTG always results in the

inequality
U(r) ≤ V (r) . (21)
12. Gravitational effects in the Solar
system
Before proceeding to examine such effects, we shall first dwell

upon certain general assertions of RTG and GRT, which
are explicitly manifested in calculations of gravitational ef-
fects. The RTG equations (5.19) and (5.20) are universally
covariant under arbitrary transformations of coordinates and
form-invariant under the Lorentz transformations. In other
words, the situation in RTG is the same as in electrody-
namics. If in two inertial reference systems the respective
distributions of matter in Galilean coordinates, Tµν [x, gαβ (x)]
and Tµν [x! , gαβ (x! )], are identical, then by virtue of the form-
invariance of the equations relative to the Lorentz transfor-
mations we obtain identical equations, which in identical con-
ditions of the problem provide for the relativity principle to
be satisfied. On the other hand, if in a certain inertial refer-
ence system and for a given distribution of matter Tµν (x) the
solution we have is gµν (x), then applying the Lorentz trans-
formations to another inertial reference system we obtain the
!
metric gµν (x! ), but it corresponds to the distribution of matter
!
Tµν (x! ). Owing to the equations being form-invariant under
the Lorentz transformations we can return to the initial vari-
!
ables x and obtain a new solution gµν (x) corresponding to the
!
distribution of matter Tµν (x). This means that a unique cor-
respondence exists between the distribution of matter and the
metric. When the distribution of matter changes, the metric
changes also. An essential point in RTG is the presence of
the metric of Minkowski space in the equations. Precisely this
circumstance permits performing a comparison of the motion
of matter in a gravitational field with the motion of matter in
the absence of any gravitational field.
154
In GRT the situation is quite different. The GRT equa-
tions outside matter are form-invariant relative to arbitrary
transformations of coordinates, and therefore, if for the distri-
bution of matter Tµν (x) the solution we have is gµν (x), then by
transforming coordinates, so that in the region of matter they
coincide with the initial ones, and outside matter differ from
them, our solution in the new coordinates will have the form
!
gµν (x! ). Owing to the equations being form-invariant outside
matter, we can go back to the initial variables x, and, conse-
!
quently, obtain a new solution gµν (x) for the same distribution
of matter, Tµν (x). To these two metrics (any amount of met-
rics can be constructed) there correspond differing intervals:
ds21 = gµν (x)dxµ dxν
ds22 = gµν
!
(x)dxµ dxν .
Which interval must be chosen? The point is that the
geodesic lines of these intervals differ from each other. In this
connection, attempts are made to identify the gravitational
field in GRT with the class of equivalent diffeomorphic metrics
!
gµν (x), gµν (x)..., obtained with the aid of transformations of
coordinates. From the point of view of mathematics this is
obvious, but what about the physical interpretation?
Thus, there exists a fundamental difference between the
conclusions in RTG and GRT, and its essence consists in that
the RTG equations are not form-invariant with respect to ar-
bitrary coordinate transformations, while the GRT equations
outside matter are form-invariant relative to such transforma-
tions. The RTG equations are only form-invariant relative to
such transformations of coordinates that leave the Minkowski
metric γµν (x) form-invariant. Hence, for example, follows the
form-invariance of equations with respect to the Lorentz trans-
formations.
155
!
The issue of the multiplicity of solutions gµν (x), gµν (x)...
worried A. Einstein seriously, and he discussed this issue in
detail in 1913–1914 in four articles [29] and arrived at the
conclusion that the choice of coordinate reference systems is
limited, since he considered that from the universal covari-
ance for one and the same distribution of matter Tµν (x) there
arises a whole set of metrics, which is physically inadmissible.
However, the main reason for ambiguity is not related to the
general covariance, it is related to the form-invariance of GRT
equations outside matter with respect to arbitrary transfor-
mations of coordinates. To remove this ambiguity, there is no
need to renounce general covariance, since it is not the cause,
but it is necessary to restrict the form-invariance of equations
in accordance with the relativity principle. Precisely this is
done in RTG on the basis of the field approach. A simple ex-
ample from electrodynamics can be presented. Assume that
for a current jµ (x) we have the solution Aµ (x). Upon perform-
ing transformations to the new variables x! coinciding with the
initial variables x in the region of the distribution of current
jµ (x) and differing from them in the region outside the cur-
rent, our solution assumes the form A!µ (x! ). But it is absolutely
obvious that A!µ (x) will not be a solution of the equations of
electrodynamics in the coordinates x, since the equations of
electrodynamics are not form-invariant with respect to arbi-
trary transformations of coordinates.
This means that in electrodynamics for one and the same
distribution of current jµ (x) in identical conditions there ex-
ists only one distribution of the electromagnetic field E, 0 H.
0
If in GRT the distribution of matter, determined by the ten-
sors Tµν (x) and Tµν (x! ), is the same in two arbitrary reference
systems, then the GRT equations being form-invariant out-
side matter, we can, in identical conditions, for example, have
156
identical metric coefficients gµν (x) and gµν (x! ). Precisely this
circumstance permitted A. Einstein to put forward the gen-
eral relativity principle for all physical processes. However,
the requirement that the metric coefficients be identical re-
sults in a strong restriction being imposed on the structure of
Riemannian space, it turns out to be a space with constant
curvature.
Since Riemannian space in GRT does not have this prop-
erty, in the general case, then the general relativity principle,
as a physical principle, is not realized in Nature. This also
follows from the equations of electrodynamics, for example,
not being form-invariant with respect to arbitrary transfor-
mations of coordinates. The relativity principle, as a physical
principle, is not related to universal covariance, but to the
form-invariance of equations and of the metric relative to the
transformations of coordinates. V.A. Fock was right, when he
wrote: “a general relativity principle, as a physical principle,
that could be valid with respect to arbitrary reference systems,
is not possible” [25]. In GRT for one and the same distribu-
tion of matter Tµν (x) there exists a whole range of solutions of
!
the GRT equations for the metric coefficients gµν (x), gµν (x), ...
Outside matter the geodesic lines for these solutions will be
different.
The issue of the multiplicity of metrics in GRT in one coor-
dinate system was widely discussed in 1921-1922 by P.Painlevé,
M.Chazy, J.Becquerel, A.Gullstrand, E.Kretschmann. The
essence of the polemic actually reduced to the question: with
which radial variable in the GRT equations is it necessary to
identify the astronomically determined distance between the
Sun and a planet? It must be noted that this arbitrariness
in the first order in the gravitational constant does not in-
fluence certain gravitational effects: the deflection of a light
157
ray, the shift of the perihelion of Mercury, the precession of a
gyroscope. However, it does, already in the first order in G,
influence the delay effect of a radiosignal.
Thus, depending on the choice of solutions in the Schwarz-
schild form or in harmonic coordinates we will obtain different
values for the delay time. We will further see that there exists
no such arbitrariness in RTG, and that effects are determined
unambiguously.
The reason for the multiplicity of metrics is not general
covariance, but the form-invariance of equations with respect
to arbitrary transformations of coordinates. There exists no
such ambiguity in RTG, since the metric gµν (x) is unambigu-
ously determined by the distribution of matter Tµν (x). In
section 11 it was shown that since the radius of a static spher-
ically symmetric body exceeds the Schwarzschild radius, then
the external solution of RTG equations in an inertial reference
system in spherical coordinates in the region (11.53) has the
form
r − MG r + MG 2
ds2 = (dx0 )2 − (dr )−
r + MG r − MG
−(r + MG)2 [(dθ)2 + sin2 Θ(dϕ)2]. (α)
Precisely such a solution in the post-Newtonian approxima-
tion yields expressions for the metric coefficients of effective
Riemannian space that coincide with the previously obtained
formulae (8.59a) applied for explanation of gravitational ef-
fects in the Solar system.
An essential point is that, when the gravitational field is
switched off (for instance, the body is removed), we necessarily
turn out to be in Minkowski space in an inertial reference
system with the metric
dσ 2 = (dx0 )2 − (dr)2 − r 2 [(dθ)2 + sin2 Θ(dϕ)2].
158
In calculating gravitational effects in the Solar system we
have to calculate the trajectory of motion in effective Rieman-
nian space, that is determined by the interval ds, and to com-
pare it with the corresponding trajectory determined by the
interval dσ. The metric of Minkowski space is present in the
RTG equations. This is precisely how the deflection angle of a
light ray and the delay time of a radiosignal, due to the influ-
ence of the gravitational field of the Sun, are determined. As
to calculating the shift in the perihelion of a planet, here it is
necessary to compare the trajectory of motion of a test body
around the Sun calculated within RTG with the trajectory
obtained from Newton’s theory of gravity. Precisely in these
calculations there exists a difference between the RTG and
GRT conclusions, since within GRT one cannot say in which
reference system (inertial or non-inertial) of Minkowski space
one happens to be, when the gravitational field is switched
off. For calculating the gravitational effect it is necessary to
compare in one coordinate system the motion along a geodesic
line in Riemannian space with motion along the geodesic line
in Minkowski space with gravity switched off. But to this end
it is necessary to know exactly both the metric gµν (x) and the
metric γµν (x).
However, in GRT, owing to the multiplicity of so-
lutions both for gµν (x) and for γµν (x) we cannot with
definiteness say which Riemannian metric gµν (x) it is
necessary to take for the chosen metric γµν (x) in order
to find the geodesic lines in Riemannian space and
in Minkowski space. This is actually the essence of
the ambiguity in predictions of gravitational effects in
GRT. Sometimes errors are avoided in GRT by considering
the initial reference system to be an inertial reference sys-
tem in Cartesian coordinates (but no such coordinates exist in
159
GRT) and then dealing with a weak gravitational field against
this background. No such difficulty exists in RTG, since for
the chosen metric γµν (x), with the aid of equations (5.19) and
(5.20), under appropriate conditions, the metric gµν (x) of ef-
fective Riemannian space is determined unambiguously, which
permits to determine unambiguously the gravitational effect.
In calculations of effects in the gravitational field of the
Sun one usually takes as the idealized model of the Sun a
static spherically symmetric body of radius R& . The general
form of the metric of Riemannian space in an inertial reference
system in spherical coordinates is
ds2 = U(r)(dx0 )2 − V (r)(dr)2 −

−W 2 (r)[(dθ)2 + sin2 Θ(dϕ)2 ]. (12.1)
In the absence of a gravitational field the metric has the form
dσ 2 = (dx0 )2 − (dr)2 − r 2 [(dθ)2 + sin2 Θ(dϕ)2 ]. (12.1a)
Substituting (12.1) and (12.1a) into equations (5.19) and

(5.20) we precisely obtain the external solution for the Sun
(α).
In section 5 it was shown that from the RTG equations
(5.19) and (5.20) follow directly the equations of motion for
matter,
∇ν T µν = 0. (12.2)
Hence it is easy to obtain the equations of motion for a test
body in a static gravitational field. The energy-momentum
tensor for matter, T µν , in this case assumes the form
dxµ
T µν = ρU µ U ν , Uµ = . (12.3)
ds
U µ ∇ν (ρU ν ) + ρU ν ∇ν U µ = 0. (12.4)
160
Multiplying this equation by Uµ and taking into account
Uµ U µ = 1 we obtain
∇ν (ρU ν ) + ρU ν Uµ ∇ν U µ = 0. (12.4a)
Since
∇ν (Uµ U µ ) = 2Uµ ∇ν U µ = 0.
from equation (12.4a) we have
∇ν (ρU ν ) = 0. (12.5)
Substituting (12.5) into (12.4) we find
U ν ∇ν U µ = 0. (12.6)
Applying the definition of a covariant derivative, equations

(12.6) may be written as
∂U µ ν
( )
µ σ dx
+ Γνσ U = 0. (12.7)
∂xν ds
Taking into account the definition of a total differential we

have
∂U µ ν
dU µ = dx . (12.8)
∂xν
On the basis of (12.8) equation (12.7) assumes the form
dU µ dxµ
+ Γµνσ U ν U σ = 0, U µ = . (12.9)
ds ds
The equation of motion of a test body, (12.9), is an equa-
tion of geodesic lines in the space with the metric gµν . The
Christoffel symbols are determined by the formula
1
Γµνσ = g µλ (∂ν gσλ + ∂σ gνλ − ∂λ gνσ ). (12.10)
2
161
On the basis of (12.1) and (12.10) it is easy to obtain the
Christoffel symbols of interest to us:
1 dW
Γ212 = , Γ233 = − sin Θ cos Θ,
W dr
(12.11)
1 dW 1 dU
Γ313 = , Γ323 = cot Θ, Γ001 = .
W dr 2U dr
Of the four equations (12.9) only three are independent,
since the following relation is valid:
gµν U µ U ν = 1. (12.12)
We shall further make use of this circumstance by choosing the

three simplest equations from (12.9). From equations (12.9)
we shall take equations
d2 x0 1 dU dx0 dr
+ = 0, (12.13)
ds2 U dr ds ds
%2
d2 Θ
$
2 dW dr dΘ dϕ
2
+ − sin Θ cos Θ = 0, (12.14)
ds W dr ds ds ds
d2 ϕ 2 dW dr dϕ dΘ dϕ
2
+ + 2 cot Θ =0 (12.15)
ds W dr ds ds ds ds
and supplement them with equation (12.12) in the form
%2 %2 %2
dx0
$ $ $
dr 2 dΘ
U −V −W −
ds ds ds
$ %2
2dϕ 2
−W sin Θ = 1. (12.16)
ds
Since the gravitational field is spherically symmetric, it
is natural to choose the reference system so as to make the
motion take place in the equatorial plane, i.e.
π
Θ= . (12.17)
2
162
In the case of our choice equation (12.14) is identically satis-
fied.
Equations (12.13) and (12.15) can be respectively written
as
dx0
( )
d
ln U = 0, (12.18)
ds ds
( )
d dϕ 2
ln W = 0. (12.19)
ds ds
Hence, we find the first integrals of motion E, J:
dx0 1 dϕ 2 J
U=√ , W =√ . (12.20)
ds E ds E
Substituting these expressions into (12.16) we obtain
%2
J2
$
1 dr
−V − = 1. (12.21)
EU ds W 2E
From the second relation (12.20) we find
√
EW 2
ds = dϕ . (12.22)
J
Passing in (12.21), with the aid of (12.22), to the variable ϕ
we obtain
$ %2
V dr 1 1 E
+ − 2 + 2 = 0. (12.23)
W4 dϕ W 2 J U J
From the first relation (12.20) we find
(ds)2 = EU 2 (dx0 )2 . (12.24)
Hence it follows that E > 0 for test bodies and E = 0 for

light.
Gravitational effects in the Solar system have been calcu-
lated within GRT by various methods. Here, we shall follow
S. Weinberg’s method of calculation [1].
163
12.1. Deflection of light rays by the Sun
Consider a photon from a distant region passing near the Sun.
What is the trajectory of the light ray? It is determined from
equation (12.23), when E = 0, and it has the form
2
3 UV
dϕ = dr 4 8. (12.25)
3
7
W2
W2 J2
−U
At the point of the light ray’s trajectory (see the figure) closest
to the Sun
dr BB
B = 0. (12.26)
dϕ r0
r l
ϕ(r)
r0
∆ϕ
Deflection of the light ray
The integral of motion J is expressed via the metric pa-

rameters U0 and W0 :
W 2 (r0 ) W2
J2 = = 0. (12.27)
U(r0 ) U0
164
Integrating (12.25) we obtain
 1/2
#∞
V
ϕ(r) = ϕ(∞) + dr  . (12.28)
 7 8
W 2 U0

r W2 W02 U
−1
The deflection angle of a light ray is
∆ϕ = 2|ϕ(r0 ) − ϕ(∞)| − π. (12.29)
Here, we have taken into account that in the absence of a

gravitational field a light ray propagates along a straight line
l, and precisely for this reason π has appeared in (12.29). For
the Sun, from the RTG equations we have
r − GM r + GM
U(r) = , V = , W 2 = (r + GM)2 . (12.30)
r + GM r − GM
For calculations it is sometimes convenient to make use of an
independent variable W :
2GM 1
U(W ) = 1 − , V (W ) = . (12.31)
W 1 − 2GM
W
In the first approximation in the gravitational constant G the

metric coefficients are
2GM 2GM
U(r) = 1 − , V (r) = 1 + ,
r r
(12.32)
2GM
& '
W 2 = r2 1 + .
r
Substituting these expressions into the integral (12.28) we ob-
tain for it the following expression:
#∞
dr
I = r0 * 7 8 C1/2 . (12.33)
4M G
r0 r r2 1 − r0
+ 4MGr − r02
165
Performing a change of variables, r = 1t , we obtain
1/r
# 0
dt
I= 6 7 82 7 82 . (12.34)
2M G 2M G
0 r0−2 1− r0
− t− r02
Making use of the tabular integral

#
dx x − 2b
6 7 82 = arcsin m + c,
b
m2 − x − 2
we find
π 2MG
+ I= . (12.35)
2 r0
On the basis of (12.29) we obtain
4M& G
∆ϕ = , (12.36)
c2 r0
taking into account
M& G
2
= 1, 475 · 105 cm, R& G = 6, 95 · 1010 cm, (12.37)
c
we find
R& " " 4M& G
∆ϕ = , = 2
= 1, 75!!. (12.38)
r0 & & R& c
Thus, the deflection of a light ray by the gravitational field
of the Sun is equal to
R&
∆ϕ = 1, 75!! · . (12.39)
r0
In calculating the deflection angle of a light ray we took into ac-
count that in the absence of a field in an inertial reference sys-
tem, a light ray travels, by virtue of the metric (12.1a), along
a straight line l. Precisely the deviation from this straight line
is the gravitational effect.
166
12.2. The delay of a radiosignal
I.I. Shapiro [43] proposed and implemented an experiment for
measurement of the time required for a radiosignal to reach the
planet Mercury and, upon reflection, to return to the Earth.
We shall calculate this time on the basis of RTG equations.
We shall pass from the independent variable ϕ to the in-
dependent variable x0 . To this end, making use of (12.22) and
(12.24), we shall obtain
J 2U 2 2
(dϕ) = 4
(dx0 )2 . (12.40)
W
With the aid of (12.40), equation (12.23) assumes the form
 1/2
#r
V
ct(r, r0 ) = dr  7 . (12.41)
 
W02
8 
U
r0 1− W2
· U0
U
Substituting expressions (12.32) into the integral (12.41), we

find
#r
rdr 2MG 2MG r0
, +
ct(r, r0 ) = - 1+ + . (12.42)
r0 r 2 − r02 r r r + r0
Applying the tabular integrals

-
#r r+ r 2 − r02
dr
- = ln ,
r0 r 2 − r02 r0
(12.43)
#r +1/2
dr r0 r − r0
,
- = ,
r0 r 2 − r 2 r + r0
0
r + r0
we obtain
-
- r+ r 2 − r02
ct(r, r0 ) = r2 − r02 + 2MG ln +
r0
+1/2
r − r0
,
+2MG . (12.44)
r + r0
167
Let re , rp be the heliocentric coordinates of the Earth and
of Mercury. Since re , rp >> r0 , then in the summands of
expressions (12.44), that contain the gravitational constant,
the influence of r0 present under the square root sign can be
neglected, which will result in
- -
ct(rp , re ) = re2 − r02 + rp2 − r02 +
2MG 4re rp 4MG
+ 2 ln 2 + 2 . (12.45)
c r0 c
We shall drop a perpendicular r⊥ from the center of the source
of the gravitational field onto the straight line connecting points
re and rp . Then, according to Pythagoras’ theorem, we have
re2 = Re2 + r⊥
2
, rp2 = Rp2 + r⊥
2
. (12.46)
In the first order in G

∆ϕ
r0 + r⊥ + Re , r02 − r⊥
2
+ Re r0 ∆ϕ, (12.47)
2
∆ϕ is the deflection angle of a light ray due to the influence
of a source of the gravitational field (see (12.36)).
5
- r0
re2 − r02 = Re 1 − ∆ϕ +
Re
∆ϕ 2MG
+ Re − r0 = Re − 2 , (12.48)
2 c
similarly
- ∆ϕ 2MG
rp2 − r02 + Rp − r0 = Rp − 2 . (12.49)
2 c
With account of (12.48) and (12.49) expression (12.45) as-
sumes the form
2MG 4re rp
ct(rp , re ) − R = ln 2 , (12.50)
c2 r0
168
here R = Re + Rp is the distance between the planets.
The delay time of a radiosignal propagating from the Earth
to Mercury and back is
4MG 4re rp
∆τ = 2[t(rp , re ) − R/c] = ln ,
c3 r0
(12.51)
re = r⊕ = 15 · 1012 cm, rp = r = 5, 8 · 1012 cm,
as r0 it is possible to take the radius of the Sun, R& :
R& = 6, 95 · 1010 cm.
Substituting these values into (12.51) and taking into account

that
4M& G
= 5, 9 · 105cm,
c2
we obtain
4M& G 4r⊕ r
∆τ = ln 2
= 219, 9 µ s. (12.52)
c3 R&
In calculating the delay effect of a radiosignal we have taken
into account that in the absence of a gravitational field a light
ray, by virtue of (12.1a), travels from point e to point p along
a straight line in an inertial reference system. Comparison
with such motion is precisely how the gravitational effect is
determined. It is precisely for this reason, that the summand
2R
c
has appeared to the left in (12.51). In observations, the
time 2Rc
is determined during a period, when the Sun moves
away from the trajectory of the light ray, so its influence is
significantly reduced.
In GRT, if the solution is sought of the Hilbert–Einstein
equations for a static spherically symmetric body of mass M,
then within one and the same coordinate system it is possible
169
to obtain the external solution for the metric, that involves
two arbitrary functions
ds2 = g00 dt2 + 2g01 dtdr+
+g11 dr 2 + g22 (dΘ)2 + g33 (dϕ)2 , (12.31a)

where
2GM
g00 = 1 − , g01 = −B(r),
W (r)
$ %2 
'−1
2GM dW
&
g11 =− 1− · − B2 ,
W dr
g22 = −W 2 (r), g33 = −W 2 sin Θ.
Thus, within one and the same coordinate system for a

body of mass M there exists an infinite number of solutions.
Here, functions B(r) and W (r) are, generally speaking, ar-
bitrary, they are not determined by GRT. P. Painlevé wrote
about all this some 80 years ago and stressed that the choice
of initial formulae is purely arbitrary. Hence it is obvious that
within GRT neither Newton’s law nor the post-Newtonian ap-
proximation (8.59a), used in GRT for explaining gravitational
effects in the Solar system, follow unambiguously from the ex-
act external solution of equations. Hence, also, it follows, for
example, that an infinitesimal period of true physical time in
GRT,  5 
2GM B(r)dr
dτ = dt 1 − −- 
W (r) 1 − 2GM W (r)
will differ depending on the choice of the arbitrary functions

W (r) and B(r). This means that for a static spherically sym-
metric body of mass M the course of physical time for one or
another process is not determined unambiguously.
170
As we can see, the situation in GRT differs totally from
the situation in electrodynamics, where the Coulomb law un-
ambiguously follows from the equations of theory. We shall
clarify the situation making use of the example of Minkowski
space. From the point of view of geometry, Minkowski space
both in an inertial and in a noninertial reference system re-
mains essentially the same, since the tensor of Riemann cur-
vature equals zero. By virtue of the form-invariance of the
Riemann tensor we will have in one and the same coordinate
!
system an infinite amount of metrics γµν (x), γµν (x)... and so
on, that reduce the Riemann tensor to zero. But, depending
on the choice of metric we will have different geodesic lines
in one and the same coordinate system, i.e. the physics will
vary. All this is obvious and well known, since the dynam-
ics in an inertial reference system differs from the dynamics
in a non-inertial reference system owing to the appearance of
forces of inertia. Precisely for this reason the choice of the
non-inertial reference system in four-dimensional Minkowski
space alters the physics. In Minkowski space, however, there
exist inertial reference systems, and observational astronomi-
cal data are referred precisely to an inertial reference system.
The choice of reference system in physical equations is based
precisely on this circumstance. In GRT Riemannian geome-
try there exist no such reference systems, so it is absolutely
unclear, which coordinates it is necessary to choose so as to
compare theoretical calculations and observational data. The
geometry does not depend on the choice of reference system
(or, in other words, in our particular case, on the choice of
functions B(r) and W (r))), it remains, like before, Rieman-
nian geometry, however, the physics changes. Naturally, it is
171
possible, in our case, to select arbitrary functions B(r) and
W (r) so as to provide for Newton’s law of gravity to be sat-
isfied, and for the post-Newtonian approximation to have the
form (8.59a). However, such a choice in GRT is, regrettably,
arbitrary, since it is not imposed by any physical conditions.
It is not possible to formulate physical requirements to be im-
posed on the behaviour of Riemannian metric, if it is not of
a field origin, because such behaviour even depends on the
choice of three-dimensional space coordinates. V.A. Fock re-
solved the issue of the choice of coordinates for island systems,
with the aid of the harmonicity conditions. But why precisely
they have to be chosen, instead of some other ones, remained
unclear.
Now, let us go back to the analysis of a concrete example
demonstrating the ambiguity of GRT in calculations of the
gravitational delay effect of a radiosignal traveling from the
Earth to Mercury and back. The predictions of theory depend
on the choice of solution. For simplicity we shall advantage of
the simplest partial case
B(r) = 0, W (r) = r + (λ + 1)M, r > R0 ,
λ is an arbitrary parameter, R0 is the radius of the body dealt

with. If an appropriate choice is made of function W (r) in the
vicinity of the body, then this solution can be made to match
the solution inside the body. If the previous calculations are
repeated, then for this metric we find
-
- r+ r 2 − r02
ct(r, r0 ) = r2 − r02 + 2MG ln +
r0
$ %, +1/2
λ r − r0
+2MG 1 + . (A)
2 r + r0
172
Comparing this expression with (12.44) we see that already
in the first order in G an ambiguity due to the influence of
the source of the gravitational field arises in GRT in the de-
scription of the delay effect of a radiosignal. On the basis of
(A) the delay time of a radiosignal traveling from the Earth
to Mercury and back, in accordance with GRT, equals the
following:
R 4MG 4re rp 2MG
, +
∆τ = 2 t(re , rp ) − = ln 2 + 3 λ. (B)
c c3 r0 c
We note that in the first order in G the physical distance
l, determined by the expression
#r
√ GM r
l= g11 dr + r − R0 + 2
ln
c R0
R0
does not depend on the parameter λ, and precisely for this

reason the first term in (B) will also not depend on λ, and,
consequently, owing to the presence of the second term, we will
have different predictions for the delay time of a radiosignal,
depending on the choice of the constant λ. Expression (B) dif-
fers essentially from the result (12.51), which follows exactly
from RTG and complies with experimental data [43]. For the
Schwarzschild solution λ = −1 the difference from (12.51) will
be 2GM
c3
, which for the Sun amounts to about 10 microseconds.
In calculations of the following gravitational effects: of the de-
flection of a light ray by the Sun, of the shift of a planet’s peri-
helion, of the precession of a gyroscope, of the shift of spectral
lines, an ambiguity also arises, but in the second order in the
gravitational constant G. All the above has been discussed in
detail with prof. Yu.M. Loskutov (see, also, monograph [10]).
173
Thus, GRT cannot, in principle, provide definite predic-
tions on gravitational effects, which is still another of its fun-
damental defects. Certain attempts to relate the gravitational
field in GRT to the equivalence class of diffeomorphic met-
rics do not remove this ambiguity, since they do not discard
the fundamental defect of GRT — the form-invariance of the
Hilbert–Einstein equations outside matter with respect to ar-
bitrary transformations of coordinates 33 . Precisely this cir-
cumstance results in the entire set of diffeomorphic metrics
arising within one coordinate system for one and the same
distribution of matter, while this, according to the Weyl–
Lorentz–Petrov theorem (see end of section 14), leads to dif-
ferent geodesic lines in identical conditions of the problem,
which is physically inadmissible. The essence of the is-
sue does not consist in the general covariance, which
must always exist, but in whether the form-invariance
of physical equations relative to arbitrary transforma-
tions of coordinates is admissible?
Since reference systems are not equivalent in the presence
of forces of inertia, no sense whatsoever can be attributed to
the form-invariance of physical equations with respect to ar-
bitrary transformations of coordinates. General covariance is
a mathematical requirement, while form-invariance has a pro-
found physical content. Actually, in all physical theories the
form-invariance of equations and of metrics holds valid rel-
ative to the Lorentz transformations — precisely this is the
essence of the relativity principle.
The non-equivalence of various reference systems is espe-
cially evident, if one considers pseudo-Euclidean geometry,
33
See, for instance: J.Stachel. Conference “Jena-1980”, 1981 (DDR).
174
for which the curvature tensor is zero. From the equality of
the curvature tensor to zero it is possible, by virtue of form-
invariance, to obtain a set of solutions for the metric tensor in
one and the same coordinate system. But it is quite obvious
that they are physically not equivalent, since some of them are
inertial reference systems, while others are non-inertial refer-
ence systems. All the above, involving significant complica-
tions, also takes place in the case of Riemannian geometry.
It is important to stress once more that in GRT,
owing to the equations being, outside the distribu-
tion of matter, form-invariant relative to arbitrary
transformations of coordinates, there arises a situa-
tion when for one and the same distribution of matter
there exists, in one and the same coordinate system,
an indefinite amount of metrics. No such situation ex-
ists in any other physical theory, since form-invariance within
them is admissible only with respect to transformations of co-
ordinates leaving the metric γµν (x) form-invariant. From this
fact, for example, the form-invariance of equations relative to
the Lorentz transformations follows.
12.3. The shift of a planet’s perihelion

Consider the motion of a test body on a solar orbit. At the
perihelion the heliocentric distance of the test body is at its
minimum and is r− , while at the aphelion it reaches its maxi-
mum and equals r+ . Since at the perihelion and at the aphe-
175
dr
lion dϕ
= 0, from equation (12.23) we obtain
1 1 E
− 2 = − 2,
W (r+ ) J U(r+ ) J
(12.53)
1 1 E
− 2 = − 2.
W (r− ) J U(r− ) J
Hence, we find
1 1
2 U+
− U−
J = 1 1 . (12.54)
W+ 2 − W− 2
Now we write equations (12.53) in another form:

$ % $ %
1 1
J 2 = W+2 − E , J 2 = W−2 −E . (12.55)
U+ U−
Hence, we obtain
W+2 2
W−
U+
− U−
E= . (12.56)
− W+2 W−2
By integration of equation (12.23) we find
#r √ , 1 1 E −1/2 dr
+
ϕ(r) = ϕ(r− ) + V − 2− 2 . (12.57)
r−
J 2U W J W2
For convenience of calculations we introduce a new indepen-

dent variable
W = r + GM. (12.58)
Substituting (12.54) and (12.56) into (12.57) and passing to
the new independent variable W we find
ϕ(W ) = ϕ(W− ) +
#W D E−1/2
W−2 [U −1 (W ) − U−−1 ] − W+2 [U −1 (W ) − U+−1 ]
+ ×
W−2 W+2 [U+−1 − U−−1 ]
W−
√
V dW
× . (12.59)
W2
176
On the basis of (12.31) we have for function U −1 (W ) in the
second order in the gravitational constant G the following:
2GM (2GM)2
U −1 (W ) = 1 + + . (12.60)
W W2
We must take into account the second order in U −1 (W ), since,
if we only consider the first order in G, then the expression
under the root sign in figure brackets will be independent of
the gravitational constant. But this means that in calculations
this circumstance results in our losing the term containing G
in the first order. We will have taken into account only the
term of the first order in G entering into function V . For the
metric coefficient of V it suffices to take into account only the
first order in G,
2GM
V (W ) = 1 + . (12.61)
W
In the approximation (12.60) the numerator of the expres-
sion under the root sign in the figure brackets of (12.59) is a
quadratic function of the variable W1 of the following form:
( $ %
1 1 1 1
2GMW− W+ (W+ − W− ) 2
− + +
W W W− W+
)
1
+ = 2GMW− W+ (W+ − W− ) ×
W− W+
$ %$ %
1 1 1 1
× − − . (12.62)
W W− W W+
The denominator of the expression under the root sign in
(12.59) is
W−2 W+2 [U+−1 − U−−1 ] = 2GMW− W+ (W− − W+ ) ×

( $ %)
1 1
× 1 + 2GM + . (12.63)
W− W+
177
Taking (12.62) and (12.63) we find the expression under
the root sign in the figure brackets (12.59),
$ %$ %
1 1 1 1 1
* 7 8C − − . (12.64)
1 + 2GM 1
+ 1 W− W W W+
W− W+
Substituting (12.61) and (12.64) into (12.59) and considering

only terms of the first order in the gravitational constant G,
we obtain
( $ % )
1 1 1
ϕ(W ) = ϕ(W− ) + 1 + + 2GM ×
2 W− W+
7 8
#W 1+ MG
dW
W
× *7 87 8C1/2 . (12.65)
1 1 1 1
W− W2 W−
− W W
− W+
For calculating the integral in (12.65) we introduce a new vari-

able ψ
$ % $ %
1 1 1 1 1 1 1
= + + − sin ψ. (12.66)
W 2 W+ W 2 W+ W−
Applying (12.66) we obtain
$ %$ %
1 1 1 1
− − =
W− W W W+
$ %2
1 1 1
= − cos2 ψ. (12.67)
4 W− W+
Upon substitution of (12.66) and (12.67) we find
7 8
#W 1+ MG
dW
W
I(W ) = *7 87 8C1/2 =
1 1 1 1
W− W2 W−
− W W
− W+
D $ %
1 1 1
= ψ + GM + ψ+
2 W− W+
$ %
1 1 1 B
+ − cos ψ } Bψ−π/2 .
B
2 W− W+
178
Hence we obtain
$ %
1 1 1
I(W+ ) = π + GM + π. (12.68)
2 W− W+
Making use of (12.68), from (12.65) we have

$ %
3 1 1
ϕ(W+ ) − ϕ(W− ) = π + πGM + . (12.69)
2 W− W+
Hence the change in angle ϕ during one revolution is

$ %
1 1 1
2|ϕ(W+ ) − ϕ(W− )| = 2π + 6πGM + . (12.70)
2 W− W+
The shift of the perihelion during one revolution will amount

to
δϕ = 2|ϕ(W+ ) − ϕ(W− )| − 2π =
$ %
1 1 1
= 6πGM + , (12.71)
2 W+ W−
or, going back to the variable determined from (12.58), we

obtain the following in the same approximation in G:
$ %
1 1 1
δϕ = 6πGM + . (12.72)
2 r+ r−
The quantities r− and r+ are expressed via the large semiaxis

a and the eccentricity e e
r± = (1 ± e)a. (12.73)
Usually a focal parameter is introduced:

$ %
1 1 1 1
= + . (12.74)
L 2 r+ r−
179
Using (12.73) we find
L = (1 − e2 )a. (12.75)

$ %
1 1 1 1
+ = . (12.76)
2 r+ r− (1 − e2 )a
Taking into account (12.76) in (12.72) we find
6πGM
δϕ = . (12.77)
c2 (1
− e2 )a
In formula (12.77) we have restored the dependence upon

the velocity of light. For Mercury
e = 0, 2056, a = 57, 91 · 1011 cm. (12.78)
Substituting these values into formula (12.77) we obtain the

shift of Mercury’s perihelion during one revolution:
δϕ = 0.1037!!. (12.79)
In one century Mercury undergoes 415 revolutions, therefore

the shift of Mercury’s perihelion in 100 years amounts to
∆ϕ = 43.03!! . (12.80)
Modern data confirm this result with an accuracy up to 1%.

Astronomers have been studying the shift of Mercury’s perihe-
lion for several centuries. In 1882 S. Newcomb established the
difference between observations and theoretical calculations,
that turned out to be 43!! per century. At present optical ob-
servations, that have been under way for over 200 years, yield
180
an uncertainty in the determination of the precession velocity
of approximately 0, 4!! per century.
To conclude this section we shall write equation (12.23) in
the variables u = W1 , W = r + GM.
%2
u2
$
du 1 E
+ − 2 + 2 = 0. (12.81)
dϕ V J UV J V
For a static spherically symmetric field, by virtue of (12.31)

we have
U = V −1 = (1 − 2GMu). (12.82)
Differentiating (12.81) with respect to ϕ and taking into ac-
count (12.82) we obtain
d2 u EGM 3GM
2
+ u = 2 2 + 2 u2 . (12.83)
dϕ J c c
Here, we have restored the dependence upon the velocity of
light. This equation differs from the equation obtained on
the basis of Newton’s theory of gravity by an additional term
3GM 2
c2
u . As we see, this term is relativistic. Precisely this term
leads to a shift in a planet’s perihelion.
Expressing the integrals of motion E and J 2 in terms of
the eccentricity and the superior semiaxis in the nonrelativistic
approximation, we have
GME 1
2 2
= . (12.84)
cJ a(1 − %2 )
Thus, in the nonrelativistic approximation we obtain the equa-

tion of Newton’s theory of gravity
d2 σ 1 1
+ σ = , σ = . (12.85)
dϕ2 a(1 − %2 ) r
181
Precisely such an expression is found in classical mechanics,
if the initial Newton equations are referred to an inertial re-
ference system. In our calculation this is natural, since the
initial RTG equations are also written in an inertial reference
system.
Comparing the motion complying with (12.83) with the
motion (12.85), we precisely determine the shift effect of the
perihelion for a single revolution of the body about the Sun. In
calculating the shift of Mercury’s perihelion and the deflection
of a light ray by the Sun, A. Einstein intuitively considered
gravity to be a weak physical field against the background of
Minkowski space. Precisely such an approach brought him to
the well-known formulae for these gravitational effects. Howe-
ver, these formulae are not unambiguous consequences of the
GRT equations. In deriving them A. Einstein rather followed
his physical intuition, than the logic of his theory. However,
upon finding these effects in 1915 he anyhow noted: “Con-
sider a material point (the Sun) at the origin of the reference
system. The gravitational field created by this material point
can be calculated from equations by successive approximations.
However, it can be assumed that for a given mass of the Sun,
gµν are not quite fully determined by equations (1) and (3).
(Here the equations Rµν = 0, given the restriction |gµν | = −1,
are intended. – A.L.) This follows from these equations being
covariant with respect to any transformations with a determi-
nant 1. Nevertheless, we, most likely, are justified in assuming
that by such transformations all these solutions can transform
into each other and that, consequently, (for given boundary
conditions) they differ from each other only formally, but not
physically. Following this conviction, I shall first restrict my-
182
self, here, to obtaining one of the solutions, without going into
the issue of whether it is the sole possible solution” 34 .
Later the issue of other possible external solutions arose
in the twenties, when the French mathematician P. Painlevé
criticized A. Einstein’s results. Following P. Painlevé, we shall
consider this issue from the point of view of the exact external
solution (12.31a) of the GRT equations for a static spherically
symmetric body.
In GRT, calculation of the shift of Mercury’s perihelion on
the basis of the exact external solution (12.31a), for a choice
of the arbitrary functions B(r) and W (r) in the simplest form
B(r) = 0, W (r) = r + (λ + 1)GM,
would lead to the following result:

(
6πGM (1 + λ)GM
δϕ = − (1 + e2 ))+
L L
e2
$ %)
9GM
+ 1+ . (12.72a)
2L 18
This expression is presented in monograph [10], therein, also,
references to original articles can be found. From formula
(12.72a) it can be seen that in GRT, also, ambiguity exists
in predicting the shift effect of Mercury’s perihelion, but it
manifests itself in the second order in G, instead of the first,
and therefore is beyond the accuracy limits of modern observa-
tional data, if one is restricted to small values of the arbitrary
parameter λ. However, from the point of view of principle it
is seen that the ambiguity is also present in the case of such
34
vol.1, art.36, p.440.
183
a choice of λ, when Newton’s law of gravity holds valid. But
from GRT it does not follow that the parameter λ should be
small. Since in the solution (12.31a) the arbitrary functions
B(r) and W (r) are not determined in GRT, therefore for a
chosen partial case the parameter λ can assume any values.
If it is chosen sufficiently large, so that the second term in
the brackets in expression (12.72a) is of the order of 10−1 , we
would arrive at a contradiction both with observational data
on the shift of Mercury’s perihelion and with Newton’s univer-
sal law of gravity. But this means, owing to the arbitrariness
of λ, that Newton’s law is not the only possible consequence
of GRT. If it were unknown to us, then from GRT, as a theo-
retical scheme, we would never obtain neither it, nor any cor-
rections to it. The maximum, that we could establish, would
be the asymptotics at infinity. All this reveals that al-
though GRT happened to be an important landmark
in gravity after the works of I. Newton, it neverthe-
less turned out to be an incomplete scheme, from the
point of view of both its physical aspects and its main
equations, applied for explaining and predicting grav-
itational phenomena.
After the sharp criticism of GRT (in the twenties) by
P. Painlevé and A. Gullstrand concerning the ambiguity in
determining gravitational effects, V.A. Fock (in the thirties)
clearly understood the essence of GRT and its not complete
definiteness. While studying island systems in the distribution
of matter in GRT, V.A. Fock added to the Hilbert–Einstein
equations harmonic coordinate conditions (actually, certain
equations were taken in Galilean coordinates of Minkowski
space, and thus departure beyond the limits of GRT was per-
184
formed) and obtained a complete set of gravitational equa-
tions. In RTG, in studying island systems of the distribution
of matter precisely such a set of equations arises in an iner-
tial reference system (in Galilean coordinates) from the least
action principle. Thus, it becomes clear, why the harmonic
conditions in Galilean (Cartesian) coordinates are universal
equations.
In studying island systems, A. Einstein and L. Infeld ap-
plied other coordinate conditions, however, in the post-Newto-
nian approximation they are close to the harmonic conditions,
and therefore within this approximation they yield the same
result. Thus, V.A. Fock’s theory of gravity permitted to un-
ambiguously determine all the effects in the Solar system. But
his approach was not consistent.
The RTG way consists in total renunciation of
A. Einstein’s ideas on inertia and gravity and return-
ing to the physical gravitational field in the spirit of
Faraday–Maxwell, exact conservation of special rela-
tivity theory, proclaiming a universal conserved quan-
tity — the energy-momentum tensor of all matter, in-
cluding the gravitational field, the source of the gravi-
tational field. Precisely such an approach leads to a new set
of equations of the theory of gravity, removes the fundamental
difficulties of GRT, discards ambiguity in the determination of
gravitational effects, predicts an another (unlike GRT) devel-
opment of the collapse and of the Universe, and at the same
time retains what is most valuable in GRT: the tensor charac-
ter of gravity and Riemannian space. But now it already stops
being the starting point and fundamental, but becomes only
effective, that arises because the energy-momentum tensor of
185
all matter, including the gravitational field, is the source of
the gravitational field.
All this is reflected in the complete set of gravitational
equations (5.19) and (5.20), that differ from the GRT equa-
tions. The effective Riemannian space, that arises in RTG
owing to the influence of the gravitational field, only has a
simple topology. This means that, in principle, no “miracles”,
that are possible in GRT owing to the complex topology of
Riemannian space, can take place in RTG.
12.4. The precession of a gyroscope

In the works of Pugh [42] and Schiff [44] a proposal was
made to put a gyroscope on an orbit around the Earth and to
examine its precession for studying the Earth’s gravitational
field and for testing general relativity theory. Precisely in this
effect, was the existence to be revealed of an inertial reference
system connected with distant stars. For simplicity, we shall
consider the gyroscope to be a pointlike test body. The equa-
tion for the angular momentum of the gyroscope, Sµ , has the
following form:
dSµ λ dxν
= Γµν Sλ . (12.86)
ds ds
In a reference system connected with the gyroscope it under-
goes no precession, which is reflected in equation (12.86). The
angular momentum of the gyroscope, J, 0 does not change in
value, below it will be shown to be expressed via the angular
momentum S 0 and the velocity 0v. In the rest frame of the test
0 and therefore
body Sµ = (0, S),
dxµ
Sµ U µ = 0, U µ = . (12.87)
ds
186
From equality (12.87) we obtain
1 dxi
S0 = − Si v i , v i = . (12.88)
c dt
Equation (12.86) for µ = i assumes the form
dSi 1
= cΓji0 Sj − Γ0i0 v j Sj + Γjik v k Sj − Γ0ik v j Sj . (12.89)
dt c
For a static spherically symmetric source of the gravita-
tional field in the linear approximation in the gravitational
constant we have
GM
g00 = 1+2Φ, g11 = g22 = g33 = 1−2Φ, Φ = − 2 . (12.90)
c r
Applying these expressions we calculate the Christoffel sym-
bols
∂Φ
Γji0 = 0, Γ0ik = 0, Γ0i0 = i ,
∂x
(12.91)
j ∂Φ ∂Φ ∂Φ
Γik = k δij − i δjk + j δik .
∂x ∂x ∂x
Substituting these expressions into equation (12.89) we obtain
0
dS 0
= −2(0v S)∇Φ − (0v ∇Φ)S0 + (S∇Φ)0
0 v. (12.92)
dt
The following expression will be the integral of motion of this
equation:
J02 = S
0 2 + 2ΦS0 2 − (0v S)
0 2. (12.93)
This is readily verified by differentiating it with respect to
time:
dJ0 0
0 dS + 4ΦS
0
0 dS +
2J0 = 2S
dt  dt dt 
 dS0 
0 (S∇Φ)
+2(0v S) 0 + 0v   . (12.94)
 dt 
187
Retaining the principal terms in this expression, we have
dJ0 0
0 dS + 2(0v S)(
2J0 = 2S 0 S∇Φ).
0 (12.95)
dt dt
0 and retaining the principal
Multiplying equation (12.92) by S
terms, we obtain
0
0 dS = −(0v S)(
S 0 S∇Φ).
0 (12.96)
dt
dJ0
J0 = 0. (12.97)
dt
Thus, we have established that expression (12.93) is an
integral of motion of equation (12.92). On the basis of (12.93)
0 Within the limits of
it is possible to construct the vector J.
accuracy it has the form
0 − 1 0v (0vS).
J0 = (1 + Φ)S 0 (12.98)
2
Differentiating (12.98) with respect to time, within the limits
of our accuracy, we obtain
dJ0 0 J],
0 Ω0 = − 3 [0v , ∇Φ].
= [Ω, (12.99)
dt 2
0 while remaining the same in absolute value,
The vector J,
undergoes precession with a velocity |Ω|0 about the direction
0 At present such an experiment is at the stage
of vector Ω.
of preparation. The precession of a gyroscope, determined by
formula (12.99), shows that a reference system, connected with
a gyroscope undergoing free motion, is not inertial. From the
point of view of RTG this is obvious, since the motion of a
188
gyroscope in the gravitational field represents an accelerated
motion with respect to the inertial reference system related to
distant stars. Precisely for this reason, the reference system
connected with the gyroscope will be non-inertial, which is
what causes precession of the gyroscope. In GRT a reference
system connected with a gyroscope, undergoing free motion,
is considered inertial. But then it is absolutely unclear, why
this inertial reference system rotates with an angular velocity
0 relative to distant stars.
of |Ω|
12.5. The gravitational shift

of spectral lines
Consider a stationary gravitational field, i.e. when the metric
coefficients are independent of time. Let radiation be emitted
from point e by the source, and let it be received at point p by
a receiver. If the source emits radiation during a time interval
(dt)e , then the receiver will also perceive it during an identical
time interval, since the gravitational field is stationary.
The proper time at point e is
√
(dτ )e = ( g00 dt)e , (12.100)
and at point p it will be

√
(dτ )p = ( g00 dt)p . (12.101)
But, since the time (dt)e = (dt)p , from formulae (12.100) and
(12.101) we obtain
2
(dτ )e 3 (g00 )e
3
=4 . (12.102)
(dτ )p (g00 )p
189
Thus, the proper time interval, during which the source
emits the signal, is not equal to the proper time interval, dur-
ing which the signal is received, since the gravitational field
differs from point e to p.
If we pass to the light frequency ω, then we obtain
2
ωe 3 (g00 )p
3
=4 . (12.103)
ωp (g00 )e
Here ωe is the frequency of light measured at the source point

e, and ωp is the frequency of the light that arrives from point
e and is measured at the receiver point p. The change in
frequency is characterized by the quantity
δω ωe − ωp
= . (12.104)
ω ωp
On the basis of (12.103) and (12.104) we find

2
δω 3 (g00 )p
3
=4 − 1. (12.105)
ω (g00 )e
For a weak gravitational field we have
g00 = 1 − 2U. (12.106)
Substituting this expression into (12.105) we obtain
δω
= Ue − Up . (12.107)
ω
If the source (for example, an atom) is in a strong gravi-
tational field, and the receiver is in a weaker field, then a red
shift is observed, and the quantity δω/ω will be positive.
13. Some other physical conclusions
of RTG
At large distances r from a static spherically symmetric body

the metric coefficients have the form
2M −mr 2M −mr
U(r) = 1 − e , V (r) = 1 + e ,
r r
M
& '
W = r 1 + e−mr .
r
We shall now deal with the problem of radiation of weak grav-
itational waves, when the graviton has mass. It has long been
well known that in linear tensor theory introduction of the
graviton mass is always accompanied by “ghosts”. However,
in refs. [15, 16, 39] it is shown that the intensity of the grav-
itational radiation of massive gravitons in nonlinear theory is
a positive definite quantity, equal to
#∞
dI 2 1
= dωω 2q{|T21|2 + |T11 −22 |2 +
dΩ πω 4
min
2
m 1 2 2 2 3m4 3 2
+ 2 (|T3 | + |T3 | ) + |T | } , (13.1)
ω 4ω 4 3
7 2
81/2
here q = 1 − mω2
.
In RTG, like in GRT, outside matter the density of the
energy-momentum tensor of the gravitational field in Rieman-
nian space equals zero:
δLg
Tgµν = −2 =0. (13.2)
δgµν
However, from expression (13.2) no absence of the gravita-

tional field follows. It is precisely in this expression, that the
191
difference between the gravitational field and other physical
fields is especially revealed. But this means that the energy
flux of the gravitational field in the theory of gravity is not
determined by the density components of the tensor Tg0i , cal-
culated with the aid of the solutions of equations (13.2), since
they are equal to zero. The problem of determining the energy
flux in the theory of gravity, unlike other theories, requires a
different approach.
Yu.M. Loskutov [15, 16, 39] finds the solution of (13.2) in
the form
Φ̃µν = χµν + ψ µν , (13.3)
where the quantities χµν and ψ µν are of the same order of
smallness, and ψ µν describes waves diverging from the source,
while χµν characterizes the background. Energy transport is
only realized by divergent waves. In ref. [15] it is shown that
the flux of gravitational energy is actually determined by the
quantity Tg0i (ψ) calculated not on the solutions of equations
(13.2) themselves, but only on that part of solutions, that
describes divergent waves ψ µν . Here it is taken into account
that gravitons do not travel in Minkowski space, like in linear
theory, but in effective Riemannian space. Therefore, in the
linear approximation the following equality is satisfied:
dxµ dxν dσ 2 − ds2

7γµν · −1 = +
ds ds ds2
1 dxα dxβ
+ − γµν Φµν + Φµν · γµα γνβ . (13.4)
2 dσ dσ
Precisely taking this circumstance into account consistently in
the course of finding the intensity has led the author of ref. [15]
to the positively definite energy flux, determined by formula
(13.1), the obtained result is of fundamental importance, since
192
it alters conventional ideas and, therefore, it necessarily re-
quires further analysis.
It must be noted that the set of gravitational equations
(5.19) and (5.20) is hyperbolic, and precisely the causality
principle provides for existence, throughout the entire space,
of a spacelike surface, which is crossed by each nonspacelike
curve in Riemannian space only once, i.e., in other words,
there exists a global Cauchy surface, precisely on which the ini-
tial physical conditions are given for one or another problem.
R. Penrose and S. Hawking [32] proved, for certain general con-
ditions, singularity existence theorems in GRT. On the basis of
equations (5.21a) outside matter, for isotropic vectors of Rie-
mannian space, by virtue of the causality conditions (6.12a),
the following inequality holds valid:
Rµν v µ v ν ≤ 0 . (13.5)
The conditions of the aforementioned theorems are contrary
to inequality (13.5), so they are not applicable in RTG.
In RTG spacelike events in the absence of a gravitational
field can never become timelike under the influence of the grav-
itational field. On the basis of the causality principle effective
Riemannian space in RTG will exhibit isotropic and timelike
geodesic completeness. In accordance with RTG, an inertial
reference system is determined by the distribution of matter
and of the gravitational field in the Universe (Mach’s princi-
ple).
In GRT the fields of inertia and of gravity are insepara-
ble. A. Einstein wrote about this: “... there exists no real
division into inertia and gravity, since the answer to the ques-
tion, of whether a body at a certain moment is exclusively un-
der the influence of inertia or under the combined influence
193
of inertia and gravity, depends on the reference system, i.e.
on the method of dealing with it” 35 . Fields of inertia satisfy
the Hilbert-Einstein equations. In RTG the gravitational field
and the fields of inertia, determined by the metric tensor of
Minkowski space, are separated, they have nothing in com-
mon. They are of different natures. The fields of inertia are
not solutions of RTG equations (5.19) and (5.20). In RTG
the fields of inertia are given by the metric tensor γµν , while
the gravitational field Φ̃µν is determined from the equations of
gravity (5.19) and (5.20).
In conclusion it must be noted that the idea that one can
arbitrarily choose both the geometry (G) and the physics (P),
since the sum (G+P) apparently seems to be the sole test
object in the experiment, is not quite correct. The choice of
pseudo-Euclidean geometry with the metric tensor γµν is dic-
tated both by fundamental physical principles — the integral
conservation laws of energy-momentum and of angular mo-
mentum, and by other physical phenomena. Thus, physics (at
the present-day stage) unambiguously determines the struc-
ture of the space-time geometry, within which all physical
fields, including the gravitational field, develop. In accor-
dance with RTG, the universal gravitational field creates effec-
tive Riemannian space with a simple topology, and Minkowski
space does not vanish, here, but is manifested in the equations
of theory and reflects a fundamental principle — the relativity
principle. Effective Riemannian space is of a field nature.
On the basis of RTG one can draw the following general

35
art.33, p.422.
194
conclusions:
Representation of the gravitational field as a physical field

possessing the energy-momentum tensor, has drastically al-
tered the general picture of gravity, earlier worked out on the
basis of GRT. First, the theory of gravity now occupies its
place in the same row as other physical theories based on the
relativity principle, i.e. the primary space is Minkowski space.
Hence, it immediately follows that for all natural phenomena,
including gravitational phenomena, there exist fundamental
physical laws of energy-momentum and angular momentum
conservation. Since a universal quantity — the conserved
energy-momentum tensor of matter (including the gravita-
tional field) is the source of the gravitational field, then there
arises effective Riemannian space-time, which is of a field na-
ture. Since the formation of effective Riemannian space-time
is due to the influence of the gravitational field, it automati-
cally has a simple topology and is described in a sole coordi-
nate system. The forces of inertia, unlike GRT, have nothing
in common with the forces of gravity, since they differ in na-
ture, the first arise owing to the choice of reference system in
Minkowski space, while the latter are due to the presence of
matter. The theory of the gravitation, like all other physical
theories (unlike GRT) satisfy the equivalence principle.
Second, the complete set of equations of the theory of gra-
vity permits to determine unambiguously gravitational effects
in the Solar system and leads to other (differing from those
of GRT) predictions both on the evolution of objects of large
mass and on the development of a homogeneous and isotropic
Universe. Theory reveals the formation of “black holes” (ob-
195
jects without material boundaries that are “cut off” from the
external world) to be impossible and predicts the existence in
the Universe of a large hidden mass of “dark” matter. From
the theory it follows that there was no Big Bang, while some
time in the past (about ten-fifteen billion years ago) there ex-
isted a state of high density and temperature, and the so-called
“expansion” of the Universe, observed by the red shift, is not
related to the relative motion of matter, but to variation in
time of the gravitational field. Matter is at rest in an inertial
reference system. The peculiar velocities of galaxies relative
to inertial reference systems are due to inhomogeneities in the
density distribution of matter, which is precisely what led to
the accumulation of matter during the period, when the Uni-
verse became transparent.
The universal integral conservation laws of energy-momen-
tum and such universal properties of matter, as, for example,
gravitational interactions, are reflected in the metric proper-
ties of space-time. While the first are embodied in the pseudo-
Euclidean geometry of space-time, the latter are reflected in
effective Riemannian geometry of space-time, that arose owing
to the presence of the gravitational field in Minkowski space.
Everything that has a character common to all matter can be
considered as a part of the structure of the effective geome-
try. But, here, Minkowski space will be present for certain,
which is precisely what leads to the integral conservation laws
of energy-momentum and angular momentum, and, also, pro-
vides for the equivalence principle to be satisfied, when the
gravitational field, as well as other fields, are switched off.
Appendix A
Let us establish the relation

δL δL ∂gαβ δ%L
= · + , (A.1)
δγµν δgαβ ∂γµν δγµν
here $ %
δL ∂L ∂L
= − ∂σ , (A.2)
δγµν ∂γµν ∂γµν,σ
$ %
δL ∂L ∂L
= − ∂σ , (A.3)
δgµν ∂gαβ ∂gαβ,σ
the asterisk in the upper formula (A.1) indicates the varia-
tional derivative of the density of the Lagrangian with respect
to the metric γµν explicitly occurring in L. Upon differentia-
tion we obtain
∂L ∂%L ∂L ∂gαβ,σ ∂L ∂gαβ
= + · + · , (A.4)
∂γµν ∂γµν ∂gαβ,σ ∂γµν ∂gαβ ∂γµν
∂L ∂%L ∂L ∂gαβ,τ
= + · . (A.5)
∂γµν,σ ∂γµν,σ ∂gαβ,τ ∂γµν,σ
We substitute these expressions into formula (A.2):
δ%L
$ %
∂L ∂L ∂L ∂gαβ,σ
− ∂σ = + · +
∂γµν ∂γµν,σ δγµν ∂gαβ,σ ∂γµν
δ%L
$ %
∂L ∂gαβ ∂L ∂gαβ,τ
+ · − ∂σ · = +
∂gαβ ∂γµν ∂gαβ,τ ∂γµν,σ δγµν
$ %
∂L ∂gαβ ∂L ∂gαβ,τ
+ · − ∂σ · +
∂gαβ ∂γµν ∂gαβ,τ ∂γµν,σ
( $ %)
∂L ∂gαβ,σ ∂gαβ,σ
+ − ∂ρ . (A.6)
∂gαβ,σ ∂γµν ∂γµν,ρ
Now, consider expression
$ %
∂gαβ,σ ∂gαβ,σ
− ∂ρ . (A.7)
∂γµν ∂γµν,ρ
197
For this purpose we shall write the derivative gαβ,σ in the form
∂gαβ ∂gαβ
gαβ,σ = ∂σ γλω + ∂σ Φλω , (A.8)
∂γλω ∂Φλω
hence it is easy to find
∂gαβ,σ ∂gαβ ρ
= ·δ . (A.9)
∂γµν,ρ ∂γµν σ
Upon differentiating this expression we have
∂ 2 gαβ ∂ 2 gαβ
$ %
∂gαβ,σ
∂ρ = ∂σ γλω + ∂σ Φλω . (A.10)
∂γµν,ρ ∂γµν ∂γλω ∂γµν ∂Φλω
On the other hand, differentiating (A.8) with respect to

γµν we have
∂gαβ,σ ∂ 2 gαβ ∂ 2 gαβ

= ∂σ γλω + ∂σ Φλω . (A.11)
∂γµν ∂γµν ∂γλω ∂γµν ∂Φλω
Comparing (A.10) and (A.11) we find

$ %
∂gαβ,σ ∂gαβ,σ
− ∂ρ =0. (A.12)
∂γµν ∂γµν,ρ
Taking this relation into account, we obtain in (A.6)
δ%L
$ %
δL ∂L ∂gαβ ∂L ∂gαβ,τ
= + · − ∂σ · . (A.13)
δγµν δγµν ∂gαβ ∂γµν ∂gαβ,τ ∂γµν,σ
Substituting (A.9) into (A.13) we find
δ%L
( $ %)
δL ∂L ∂L ∂gαβ
= + − ∂σ , (A.14)
δγµν δγµν ∂gαβ ∂gαβ,σ ∂γµν
i.e.,
δL δ%L δL ∂gαβ
= + · . (A.15)
δγµν δγµν δgαβ ∂γµν
198
The following is calculated in a similar manner:
δL δL ∂g̃ λρ
= λρ · . (A.16)
δgαβ δg̃ ∂gαβ
Making use of (A.16), one can write expression (A.15) as fol-

lows:
δL δ%L δL ∂g̃ λρ
= + λρ · . (A.17)
δγµν δγµν δg̃ ∂γµν
Appendix B
The density of the Lagrangian of the gravitational field proper
has the form
Lg = Lg0 + Lgm , (B.1)
1 αβ 7 τ λ 8
Lg0 = − g̃ Gλα Gτ β − Gταβ Gλτλ , (B.2)
16π
m2 1 √ √ '
&
αβ
Lgm = − γαβ g̃ − −g − −γ . (B.3)
16π 2
The third-rank tensor Gταβ is
1
Gταβ = g τ λ (Dα gβλ + Dβ gαλ − Dλ gαβ ) , (B.4)
2
it is expressed via the Christoffel symbols of Riemannian space
and of Minkowski space:
Gταβ = Γταβ − γαβ
τ
. (B.5)
Let us calculate the variational derivative of Lg with re-
spect to the explicitly present metric of Minkowski space, γµν :
δ % Lg0
$ %
∂Lg0 ∂Lg0
= − ∂σ . (B.6)
δγµν ∂γµν ∂γµν,σ
For this purpose we perform certain preparatory calculations:
∂Gλαβ λ
∂γαβ 1 µ
=− = (γ λµ γαβ
ν
+ γ λν γαβ ),
∂γµν ∂γµν 2
(B.7)
∂Gλαλ ∂γ λ 1 µ
= − αλ = (γ λµ γλα
ν
+ γ λν γαλ ),
∂γµν ∂γµν 2
∂Gλαβ λ
∂γαβ 1*
=− = − γ λµ (δαν δβσ + δασ δβν )+
∂γµν,σ ∂γµν,σ 4
C
+γ λν (δαµ δβσ + δασ δβµ ) − γ λσ (δαµ δβν + δαν δβµ ) , (B.8)
∂Gλαλ ∂γ λ 1
= − αλ = − γ µν δασ .
∂γµν,σ ∂γµν,σ 2
200
Differentiating (B.2) we obtain
1 αβ ∂Gταλ λ ∂Gλτβ
(
∂Lg0
=− g̃ Gτ β + Gτλα −
∂γµν 16π ∂γµν ∂γµν
∂Gταβ λ λ
)
τ ∂Gτ λ
− G − Gαβ .
∂γµν τ λ ∂γµν
Substituting into this expression formulae (B.7) we find

∂Lg0 1 αβ τ λµ ν 1
L
=− g̃ Gλα γ γτ β + Gτλα γ λν γτµβ − Gλτλ γ τ µ γαβ
ν
−
∂γµν 16π 2
1 λ τ ν µ 1 τ λµ ν 1 τ λν µ 1 µν
M
− Gτ λ γ γαβ − Gαβ γ γτ λ − Gαβ γ γτ λ = B . (B.9)
2 2 2 32π
With the aid of the derivatives (B.8) we obtain
∂Lg0 1 σµν
= A , (B.10)
∂γµν,σ 32π
where
Aσµν = γ τ µ (Gστβ g̃ νβ + Gντ β g̃ σβ − Gλτλ g̃ σν )−
−γ µν Gσαβ g̃ αβ + γ τ ν (Gστβ g̃ µβ + Gµτ β g̃ σβ − Gλτλ g̃ σµ )+
+γ τ σ (Gλτλ g̃ µν − Gµτ β g̃ νβ − Gντ β g̃ µβ ). (B.10! )
The density of the tensor Aσµν is symmetric in indices µ

and ν. The ordinary derivative of this density can be repre-
sented in the form
∂σ Aσµν = Dσ Aσµν − γσρ

µ
Aσρν − γσρ
ν
Aσµρ .
Substituting into (B.6) expressions (B.9) and (B.10) we find

δ % Lg0 1 µν 1
= B − Dσ Aσµν +
δγµν 32π 32π
1 µ σρν 1 ν σµρ
+ γσρ A + γ A . (B.11)
32π 32π σρ
201
Now, we write the density of the tensor Aσρν in the form
Aσρν = (Gστβ γ τ ρ g̃ νβ − Gρτβ γ τ σ g̃ νβ ) + (Gντ β γ τ ρ g̃ σβ − Gντ β γ τ σ g̃ ρβ )−

−(Gλτλ γ τ ρ g̃ σν − Gλτλ γ τ σ g̃ ρν ) + Gστβ γ τ ν g̃ ρβ + Gρτβ γ τ ν g̃ σβ −
−Gλτλ γ τ ν g̃ σρ − Gσαβ γ ρν g̃ αβ ,
in the brackets, terms antisymmetric in indices σ and ρ have

been formed. Writing in such a way facilitates finding the
µ
expression for the quantity γσρ Aσρν , since the terms antisym-
metric in indices σ and ρ vanish automatically, here.
µ
γσρ Aσρν = 2Gστβ γσρ
µ τ ν ρβ
γ g̃ −
−Gλτλ γσρ
µ τ ν σρ
γ g̃ − Gσαβ γσρ
µ νρ αβ
γ g̃ . (B.12)
Representing in a similar manner Aσµρ as
Aσµρ = (Gστβ γ τ ρ g̃ µβ −Gρτβ γ τ σ g̃ µβ ) + (Gµτ β γ τ ρ g̃ σβ −Gµτ β γ τ σ g̃ ρβ ) +

+(Gλτλ γ τ σ g̃ µρ − Gλτλ γ τ ρ g̃ σµ ) + Gστβ γ τ µ g̃ ρβ + Gρτβ γ τ µ g̃ σβ −
−Gλτλ γ τ µ g̃ σρ − Gσαβ γ µρ g̃ αβ ,
where again in the brackets terms antisymmetric in the indices

σ and ρ are formed, we obtain
ν
γσρ Aσµρ = 2Gστβ γσρ
ν τ µ ρβ
γ g̃ −
−Gλτλ γσρ
ν τ µ σρ
γ g̃ − Gσαβ γσρ
ν µρ αβ
γ g̃ . (B.13)
Summing (B.12) and (B.13) one readily verifies the following

equality:
µ
γσρ Aσρν + γσρ
ν
Aσµρ = −B µν . (B.14)
Taking this equality into account we write expression (B.11)
in the form
δLg0 1
=− Dσ Aσµν . (B.15)
δγµν 32π
202
Taking into account the equalities
1 √ √
Gλτλ = g λρ Dτ gλρ , Dτ −g = −gGλτλ ,
2
we find
Gστβ g̃ νβ + Gντ β g̃ σβ − Gλτλ g̃ σν = −Dτ g̃ νσ ,

Gστβ g̃ µβ + Gµτ β g̃ σβ − Gλτλ g̃ σµ = −Dτ g̃ µσ (B.16)
Gντ β g̃ µβ + Gµτ β g̃ νβ − Gλτλ g̃ µν µν
= −Dτ g̃ .
Substituting these expressions into (B.10! ) we obtain
Aσµν = γ τ σ Dτ g̃ µν + γ µν Dτ g̃ τ σ − γ τ µ Dτ g̃ νσ − γ τ ν Dτ g̃ µσ .
Substituting this expression into (B.15) we find
δ % Lg0 1 µν
= J , (B.17)
δγµν 32π
where J µν = −Dσ Dτ (γ τ σ g̃ µν + γ µν g̃ τ σ − γ τ µ g̃ νσ − γ τ ν g̃ µσ ) .
On the basis of (B.3) we have
δ % Lgm m2 µν m2 µν
=− (g̃ − γ̃ µν ) = − Φ̃ . (B.18)
δγµν 32π 32π
Thus, taking into account (B.1) and applying (B.17) and

(B.18) we find
δ % Lg 1
= (J µν − m2 Φ̃µν ) , (B.19)
δγµν 32π
and, consequently,
δ % Lg 1
−2 = (−J µν + m2 Φ̃µν ) . (B.20)
δγµν 16π
203
Appendix B∗
In this Appendix we shall make use of expressions (B.2) and

(B.3) for the density of the Lagrangian Lg0 and Lgm in order
to establish the following equalities:
δLg0 1 δLgm m2
=− Rαβ , = (gαβ − γαβ ), (B∗.1)
δg̃ αβ 16π δg̃ αβ 32π
δLg0 δLgm
here, by definition, the tensors δg̃ αβ
, δg̃ αβ
are equal to
δLg0 ∂Lg0 ∂Lg0 δLgm ∂Lgm ∂Lgm

αβ
= αβ − ∂σ αβ , αβ
= αβ
− ∂σ αβ . (B∗.2)
δg̃ ∂g̃ ∂g̃,σ δg̃ ∂g̃ ∂g̃,σ
The tensor relations (B∗.1) are most readily established in a
local Riemann reference system, where the derivatives of the
components of the metric tensor gµν with respect to the co-
ordinates are zero and, consequently, the Christoffel symbols
Γλµν are also zero.
On the basis of the formula
∂Γτλα 1
= − (g µτ Γναλ + g ντ Γµαλ ) . (B∗.3)
∂gµν 2
it is easy to establish that in the indicated reference system
the following equality holds valid:
√
∂Lg0 −g αµ βν 1 µν αβ
& '
= g g − g g )×
∂gµν 16π 2
7 8
τ
× γλα γτλβ − γαβ
τ
γτλλ . (B∗.4)
Here, γτλβ are the Christoffel symbols of Minkowski space.

Making use of expression
∂Γταλ 1
= {g τ µ (δαν δλσ + δλν δασ ) + g τ ν (δαµ δλσ +
∂gµν,σ 4
+δλµ δασ ) − g τ σ (δαµ δλν + δλµ δαν )} , (B∗.5)
204
we obtain
√
∂Lg0 −g 1
,
− = (g αµ g βν − g µν g αβ )(Γσαβ − γαβ
σ
)−
∂gµν,σ 16π 2
(B∗.6)
1
+
− (g αµ g σν − g µν g ασ )(Γλαλ − γαλ
λ
) .
2
Hence in a local Riemann reference system we find
√
∂Lg0 −g αµ βν 1 µν αβ
& '
− ∂σ = g g − g g ·
∂gµν,σ 16π 2
* C (B∗.7)
× (∂σ Γσαβ − ∂β Γλαλ ) − σ
(∂σ γαβ − λ
∂β γαλ ) .
On the basis of (B∗.4) and (B∗.7) the tensor (B∗.2) in the local
Riemann reference system is
√
δLg0 ∂Lg0 ∂Lg0 −g 7 αµ βν
= − ∂σ = g g −
δgµν ∂gµν ∂gµν,σ 16π
(B∗.8)
1
'
− g µν g αβ (∂σ Γσαβ − ∂β Γλαλ − Rαβ (γ)).
2
In a local Riemann reference system the second-rank cur-
vature tensor of Riemann space, Rαβ (g), has the form
Rαβ (g) = ∂σ Γσαβ − ∂β Γλαλ . (B∗.9)
In expression (B∗.8), the second-rank tensor Rαβ (γ) is

σ λ τ
Rαβ (γ) = ∂σ γαβ − ∂β γαλ + γαβ γτλλ − γλα
τ
γτλβ .
In Minkowski space with the metric γµν and Christoffel sym-

σ
bols γµν the tensor Rαβ (γ) equals zero. Taking into account
∗
(B .9) and, also, that the tensor Rαβ (γ) equals zero, the tensor
relation (B∗.8) assumes the form
√
∂Lg0 −g 1
& '
= Rµν − g µν R . (B∗.10)
∂gµν 16π 2
205
This equality has been established in a local Riemann reference
system, but by virtue of its tensor character it holds valid in
any reference system.
Applying relation
dg = −ggαβ dg αβ , (B∗.11)
we find
∂g
= gg̃αβ . (B∗.12)
∂g̃ αβ
On the basis of equality
√
g̃ µσ gσν = δνµ −g (B∗.13)
it is easy to obtain the following relation:

∂gλν 1 1 1
L M
αβ
=√ − (gλα gνβ + gλβ gνα ) + gαβ gνλ . (B∗.14)
∂g̃ −g 2 2
Since on the basis of Appendix A the following equality is

valid:
δLg0 δLg0 ∂gλν
αβ
= · , (B∗.15)
δg̃ δgλν ∂g̃ αβ
then, upon applying expressions (B∗.10) and (B∗.14) we find
δLg0 1
αβ
=− Rαβ . (B∗.16)
δg̃ 16π
In a similar manner we have
δLgm ∂Lgm m2
= = (gαβ − γαβ ). (B∗.17)
δg̃ αβ ∂g̃ αβ 32π
Adding up expressions (B∗.16) and (B∗.17) we obtain
δLg 1 m2
= − Rαβ + (gαβ − γαβ ). (B∗.18)
δg̃ αβ 16π 32π
206
It is also easy to obtain the following relation:
δLM δLM ∂gλν 1 λν ∂gλν
= · = − T =
δg̃ αβ δgλν ∂g̃ αβ 2 ∂g̃ αβ
1 1
= √ (Tαβ − gαβ T ). (B∗.19)
2 −g 2
Here T λν = −2 δL M
δgλν
is the density of the energy-momentum
tensor of matter in effective Riemannian space. At last we
present the following relations:
$5 %
√ g √
Dν −g = ∂ν = −gGλνλ ,
γ (B∗.20)
√ √ λ √ √ λ
∂ν −g = −gΓνλ , ∂ν −γ = −γγνλ ,
that are applied in obtaining equality (5.15).

Appendix C
For any given density of the Lagrangian L, in the case of an

infinitesimal change in the coordinates, the variation of the
action #
S = Ld4 x
will be zero. We shall calculate the variation of the action of
the density of the Lagrangian LM
#
SM = LM (g̃ µν , ΦA )d4 x
of matter and establish a strong identity. In the case of the

following transformation of the coordinates:
x!µ = xµ + ξ µ (x) , (C.1)
where ξ µ (x) is an infinitesimal displacement four-vector, the

variation of the action under transformation of the coordinates
equals
$ %
δLM δLM
#
4
δc SM = dx µν
δL g̃ µν + δL ΦA + div = 0. (C.2)
δg̃ δΦA
In this expression div stands for the divergence terms, which
are not essential for our purposes.
The Euler variation is defined as usual:
δL ∂L ∂L ∂L
≡ − ∂µ + ∂µ ∂ν .
δΦ ∂Φ ∂(∂µ Φ) ∂(∂µ ∂ν Φ)
The Lie variations δL g̃ µν , δL ΦA are readily calculated under
changes of the coordinates, if the transformation law of the
quantities g µν , ΦA is applied:
δL g̃ µν = g̃ λµ Dλ ξ ν + g̃ λν Dλ ξ µ − Dλ (ξ λ g̃ µν ),
(C.3)
λ B;λ
δL ΦA = −ξ Dλ ΦA + FA;σ ΦB Dλ ξ σ ,
208
Dλ are covariant derivatives in Minkowski space. Substituting
these expressions into (C.2) and integrating by parts we obtain
D ( $ % $ %
#
δLM δLM
δSM = d4 x −ξ λ Dα 2 λν g̃ αν − Dλ g̃ αβ +
δg̃ δg̃ αβ
$ % ) E
δLM B;σ δLM
+ Dσ FA;λ ΦB + Dλ ΦA + div = 0 . (C.4)
δΦA δΦA
Owing to the arbitrariness of vector ξ λ , we derive from this

equality a strong identity, which is valid independently of
whether the equations of motion for the fields are satisfied
or not. It has the form
$ % $ %
δLM δLM
Dα 2 λν g̃ αν − Dλ g̃ αβ =
δg̃ δg̃ αβ
$ %
δLM B;σ δLM
= −Dσ FA;λ ΦB − Dλ ΦA . (C.5)
δΦA δΦA
We now introduce the notation
δLM δLM
Tµν = 2 µν
, T µν = −2 = g µα g νβ Tαβ ,
δg δgµν
δLM
T = T µν gµν , T̃µν = 2 µν , (C.6)
δg̃
δLM
T̃ µν = −2 = g̃ µα g̃ νβ T̃αβ , T̃ = T̃ αβ g̃αβ .
δg̃µν
Taking into account this notation, one can write the left-hand
side of identity (C.5) as
1 1
Dα (T̃λν g̃ αν ) − g̃ αβ Dλ T̃αβ = ∂α (T̃λν g̃ αν ) − g̃ αβ ∂λ T̃αβ .
2 2
The right-hand side of this equation is readily reduced to the
form
1 1
& '
αν
∂α (T̃λν g̃ ) − g̃ αβ ∂λ T̃αβ = g̃λν ∇α T̃ αν − g̃ αν T̃ , (C.7)
2 2
209
where ∇α is the covariant derivative in Riemannian space.
We shall now represent the expression under the sign of the
covariant derivative ∇α in terms of the density of the tensor
T αν . To this end we take advantage of formula (A.16):
δLM δLM ∂g̃ αβ
= αβ · , (C.8)
δgµν δg̃ ∂gµν
where
∂g̃ αβ √ ∂g αβ 1 ∂g αβ
= −g − √ · g . (C.9)
∂gµν ∂gµν 2 −g ∂gµν
Using the relations
g αβ gβσ = δσα ,
we find
∂g αβ 1
= − (g αµ g νβ + g αν g µβ ) . (C.10)
∂gµν 2
Applying the rule for differentiating determinants we find
dg = gg µν dgµν , (C.11)
from which we find

∂g
= gg µν . (C.12)
∂gµν
Substituting expressions (C.10) and (C.12) into (C.9) we ob-
tain
∂g̃ αβ 1√
=− −g[g αµ g βν + g αν g βµ − g µν g αβ ] . (C.13)
∂gµν 2
Using this relation in (C.8) we find
$ %
δLM √ δLM αµ βν 1 δLM αβ µν
= −g g g − g g . (C.14)
δgµν δg̃ αβ 2 δg̃ αβ
With account of notation (C.6) this expression can be written
in the form
√ 1
−gT µν = T̃ µν − g̃ µν T̃ . (C.15)
2
210
On the basis of equality (C.15), the strong identity (C.5)
assumes, with account of (C.7), the following form
$ %
αν δLM B;σ δLM
gλν ∇α T = −Dσ FA;λ ΦB − Dλ ΦA , or
δΦA δΦA
$ % (C.16)
δLM B;σ δLM
∇α Tλα = −Dσ FA;λ ΦB − Dλ ΦA .
δΦA δΦA
Appendix D
A second-rank curvature tensor Rµν can be written in the form

1 1
Rµν = [g̃ αβ (g̃µκ g̃νρ − g̃µν g̃κρ )Dα Dβ g̃ κρ −
2 2
1
−g̃νρ Dκ Dµ g̃ κρ − g̃µκ Dν Dρ g̃ κρ] + g̃νω g̃ρτ Dµ g̃ κρDκ g̃ ωτ +
2
1 1
+ g̃µω g̃ρτ Dν g̃ κρDκ g̃ ωτ − g̃µω g̃νρ Dτ g̃ ωκ Dκ g̃ ρτ −
2 2
1 1
− (g̃ωρ g̃κτ − g̃ωτ g̃κρ )Dµ g̃ κρDν g̃ ωτ −
4 2
1 αβ 1
− g̃ g̃ρτ (g̃µκ g̃νω − g̃µν g̃κω )Dα g̃ κρDβ g̃ ωτ . (D.1)
2 2
Raising the indices by multiplying by g -µg λν and taking into
account the equation
Dµ g̃ µν = 0 , (D.2)
we obtain
1 1
−gR-λ = g̃ αβ Dα Dβ g̃ -λ − g̃ -λg̃κρg̃ αβ Dα Dβ g̃ κρ +
2 4
1
+ g̃ρτ g̃ -µ Dµ g̃ κρ Dκ g̃ λτ +
2
1 1
+ g̃ρτ g̃ λν Dν g̃ κρ Dκ g̃ -τ − Dτ g̃ -κ Dκ g̃ λτ −
2 2
1 1
− (g̃ωρ g̃κτ − g̃ωτ g̃κρ )g̃ -µg̃ λν Dµ g̃ κρDν g̃ ωτ −
4 2
1 1
− g̃ρτ g̃ αβ Dα g̃ -ρDβ g̃ λτ+ g̃ρτ g̃ -λg̃κω g̃ αβ Dα g̃ κρ Dβ g̃ ωτ. (D.3)
2 4
Hence, we find
1 1
−gR = g-λg̃ αβ Dα Dβ g̃ -λ − gκρg̃ αβ Dα Dβ g̃ κρ + gρτ Dµ g̃ κρ Dκ g̃ µτ−
2 2
1 1 √
− (g̃ωρ g̃κτ − g̃ωτ g̃κρ) −gg̃ µν Dµ g̃ κρDν g̃ ωτ −
4 2
1
− g̃ρτ g̃ αβ g-λ Dα g̃ -ρDβ g̃ λτ + g̃ρτ gκω g̃ αβ Dα g̃ κρDβ g̃ ωτ . (D.4)
2
212
With the aid of expressions (D.3) and (D.4) we find
1
& '
−g( R − g -λR =
-λ
D &
2
1 1 1
'
=− g̃νσ g̃τ κ g̃νκ g̃τ σ g̃ -α g̃ λβ Dα g̃ στ Dβ g̃ νκ −
2 2 2
1 1
& '
− g̃ -λg̃ αβ g̃νσ g̃τ κ − g̃νκ g̃τ σ Dα g̃ τ σ Dβ g̃ νκ +
4 2
+g̃ g̃στ Dα g̃ Dβ g̃ − g̃ -β g̃τ σ Dα g̃ λσ Dβ g̃ ατ −
αβ -τ λσ
1
−g̃ λα g̃τ σ Dα g̃ βσ Dβ g̃ -τ + g̃ -λg̃τ σ Dα g̃ βσ Dβ g̃ ατ +
2 E
+Dα g̃ -β Dβ g̃ λα − g̃ αβ Dα Dβ g̃ -λ . (D.5)
It must be especially stressed that in finding expression (D.5)

we made use of equation (D.2). By substituting expression
(D.5) into equation (5.19) and writing the thus obtained equa-
tion in the form (8.1) we find the expression for the quantity
−16πgτg-λ :
1 1 1
−16πgτg-λ = (g̃ -αg̃ λβ − g̃ -λg̃ αβ )(g̃νσ g̃τ µ − g̃τ σ g̃νµ ) ×
2 2 2
×Dα Φ̃ Dβ Φ̃ + +g̃ g̃τ σ Dα Φ̃ Dβ Φ̃ − g̃ -β g̃τ σ Dα Φ̃λσ Dβ Φ̃ατ −
τσ µν αβ -τ λσ
1
−g̃ λα g̃τ σ Dα Φ̃βσ Dβ Φ̃-τ + g̃ -λg̃τ σ Dα Φ̃σβ Dβ Φ̃ατ +Dα Φ̃-β Dβ Φ̃λα −
&2
√ √
αβ -λ 2
−Φ̃ Dα Dβ Φ̃ − −m ( −gg̃ -λ − −γ Φ̃-λ + g̃ -αg̃ λβ γαβ −
1
'
− g̃ -λg̃ αβ γαβ ) . (D.6)
2
Appendix E
Let us write the RTG equation (5.20)
Dσ g̃ σν (y) = ∂σ g̃ σν (y) + γαβ

ν
(y)g̃ αβ (y) = 0 (Σ)
in a somewhat different form. For this purpose, making use of

the definition of a Christoffel symbol,
1
Γναβ (y) = g νσ (∂α gσβ + ∂β gσα − ∂σ gαβ ), (E.1)
2
we find
√ &
1
'
Γναβ g̃ αβ (y) = νσ αβ
−g g g ∂α gσβ − g νσ g αβ ∂σ gαβ . (E.2)
2
Taking into account the equalities
1 1 -
Γλσλ = g αβ ∂σ gαβ = √ ∂σ −g(y),
2 −g
∂α g αν = −g νσ g αβ ∂α gσβ (E.3)
we rewrite (E.2) as
√ √ ∂g̃ σν
Γναβ (y)g̃ αβ (y) = − −g∂σ g σν − g νσ ∂σ −g = − σ . (E.4)
∂y
With account of this equality the initial equation (Σ) assumes
the form
(Γναβ (y) − γαβ
ν
(y))g αβ (y) = 0. (E.5)
If we pass from coordinates “y” to other curvilinear coor-
dinates “z”, then the Christoffel symbols assume the form
∂y λ ∂z α ∂z β σ
Γλµν (y) = · Γ (z) +
∂z σ ∂y µ ∂y ν αβ
∂2zσ ∂y λ
+ µ ν · σ. (E.6)
∂y ∂y ∂z
214
Applying this expression we find
∂y λ
(
= σ Γσαβ (z)g αβ (z)+
Γλµν (y)g µν (y)
∂z
∂ 2 z σ ∂y µ ∂y ν αβ
)
+ µ ν α · β g (z). (E.7)
∂y ∂y ∂z ∂z
On the basis of (E.4) we write expression (E.7) in the form
λ
$ %
1 ∂ µσ ∂y
Γλµν (y)g µν (y)
= −√ g̃ +
−g ∂z µ ∂z σ
∂2yλ ∂y λ ∂ 2 z σ ∂y µ ∂y ν
+g µσ µ σ + σ · µ ν α β g αβ (z). (E.8)
∂z ∂z ∂z ∂y ∂y ∂z ∂z
Upon differentiating equality
∂z σ ∂y µ
· = δασ (E.9)
∂y µ ∂z α
with respect to the variable z β we obtain
∂ 2 z σ ∂y µ ∂y ν ∂z σ ∂2yµ
· = − · . (E.10)
∂y µ ∂y ν ∂z α ∂z β ∂y µ ∂z α ∂z β
Taking into account this equality, in the third term of (E.8)
we find
λ
$ %
1 ∂ νσ ∂y
Γλµν (y)g µν (y) = −- ν
g̃ . (E.11)
−g(z) ∂z ∂z σ
Substituting this expression into (E.5) we obtain
y λ = −γαβ
λ
(y)g αβ (y), (E.12)
where denotes the operator

$ %
∂1 νσ ∂
=- ν
g̃ . (E.13)
−g(z) ∂z ∂z σ
14. Elements of tensor analysis and
of Riemannian geometry
Consider a certain coordinate system xα , i = 1, ...n to be de-

fined in n-dimensional space. Instead of this coordinate system
one may also choose another one defined by expression
x!α = f (xα ), α = 1, ...n. (14.1)
These functions must be continuous and have continuous par-

tial derivatives of order N. If the transformation Jacobian at
each point,
B ∂f α B
B B
J = det B β B , (14.2)
B B
B ∂x B
differs from zero, then in this condition the variables x!α will
be independent, and, consequently, the initial variables xα can
be unambiguously expressed in terms of the new ones, x!α :
xα = ϕ(x!α ). (14.3)
Physical quantities must not depend on the choice of coor-

dinate system, and therefore they must be expressed in terms
of geometrical objects. The simplest geometrical object is a
scalar, that transforms in transition to the new coordinates as
follows:
Φ! (x! ) = Φ(x(x! )). (14.4)
The gradient of a scalar function Φ(x) transforms in accor-
dance with the rule for the differentiation of composite func-
tions,
∂Φ! (x! ) ∂Φ ∂xβ
= · . (14.5)
∂x!α ∂xβ ∂x!α
216
Here, summation is performed over identical indices β. The set
of functions transforming under coordinate transformations by
the rule (14.5) is termed the covariant vector
∂xβ
A!α (x! ) = Aβ (x) !α . (14.6)
∂x
Correspondingly the quantity Bµν is a covariant second-rank
tensor, that transforms by the rule
! ∂xα ∂xβ
Bµν (x! ) = Bαβ (x) · (14.7)
∂x!µ ∂x!ν
and so on.
We shall now pass to another group of geometric objects.
Consider transformation of the differential of coordinates
∂x!µ α
dx!µ = dx . (14.8)
∂xα
A set of functions transforming under coordinate transforma-
tions by the rule (14.8) has been termed a contravariant vector,
∂x!µ α
A!µ (x! ) = A (x), (14.9)
∂xα
correspondingly, the quantity B µν a contravariant second-rank
tensor transforming by the rule
∂x!µ ∂x!ν αβ
B !µν (x! ) = B (x) (14.10)
∂xα ∂xβ
and so on. Expressions (14.6), (14.7), (14.9) and (14.10) per-
mit to write the transformation law of tensors of any form.
For example,
∂x!µ ∂xβ α
Bν!µ (x! ) = · B (x) (14.11)
∂xα ∂x!ν β
217
From the transformational properties of a tensor it follows
that, if all its components are equal to zero in one coordinate
system, then they equal zero in another coordinate system,
also. It is readily verified that the transformations of covariant
and contravariant quantities exhibit the group property. For
example:
∂x!µ α ∂x!!ν !µ !
A!µ = A (x), A!!ν
(x!!
) = A (x ),
∂xα ∂x!µ
(14.12)
∂x!!ν ∂x!µ α ∂x!!ν α
A!!ν (x!! ) = · A (x) = A (x).
∂x!µ ∂xα ∂xα
Now, let us pass to tensor algebra. Here, four operations are
possible: addition, multiplication, convolution, and permuta-
tion of indices.
Addition and subtraction of tensors
If we have tensors of identical structure, i.e. that have the

same number of contravariant indices and the same number of
covariant indices, for example,
Aαβ αβ
µνσ , Bµνσ ,
then it is possible to form the tensor

αβ
Cµνσ = Aαβ αβ
µνσ + Bµνσ . (14.13)
Multiplication of tensors
Tensors can be multiplied independently of their structure.

For example,
αβλ
Cµνσρ = Aαβ λ
µνσ · Bρ . (14.14)
Here, both the order of multipliers and the order of indices
must be observed.
218
The convolution operation of tensors
With the aid of the Kronecker symbol

D
0 at µ -= ν
δνµ = (14.15)
1 at µ = ν,
which is a tensor, it is possible to perform the convolution
operation of indices, for example,
Aαβ ν αβ
µν · δσ = Aµσ . (14.16)
Here, on the left, summation is performed over identical in-
dices.
The permutation operation of indices
By permutation of indices of the tensor we obtain another ten-

sor, if the initial tensor was not symmetric over these indices,
for example,
µν
Bλσ = Aµν
σλ . (14.17)
With the aid of this operation, as well as addition, it is possible
to construct a tensor that is symmetric over several indices.
For example,
1
A(µν) = (Aµν + Aνµ ). (14.18)
2
It is also possible to construct a tensor, that is antisymmetric
over several indices. For example,
1
A[µν] = (Aµν − Aνµ ). (14.19)
2
Such an operation is called antisymmetrization.
Riemannian geometry
A Riemannian space Vn is a real differentiable manifold, at
each point of which there is given the field of a tensor
gµν (x) = gµν (x1 , . . . , xn ) (14.20)
219
twice covariant, symmetric and nondegenerate
gµν = gνµ , g = det |gµν | =
- 0. (14.21)
The tensor gµν is called a metric tensor of Riemannian space.
The functions gµν are continuous and differentiable with re-
spect to all variables x1 , . . . , xn up to the n-th order.
With the aid of the metric tensor in Riemannian space it
is possible to introduce an invariant differential form termed
an interval
(ds)2 = gµν (x)dxµ dxν . (14.22)
With the aid of coordinate transformations this form at any
fixed point can be reduced to a diagonal form. Here, in the
general case, the diagonal components of the matrix gµν will
not all be positive. But, by virtue of the law of inertia for
quadratic forms the difference between the amounts of pos-
itive and of negative diagonal components will be constant.
This difference is called the signature of a metric tensor. In an
arbitrary Riemannian space Vn the interval will exhibit alter-
nating signs. We shall further call it timelike, if ds2 > 0, space-
like, if ds2 < 0, isotropic, if ds2 = 0. These terms originated
within special relativity theory, where space and time form
a unique manifold, while the interval in Cartesian (Galilean)
coordinates has the form
dσ 2 = (dx0 )2 − (dx1 )2 − (dx2 )2 − (dx3 )2 . (14.23)
In arbitrary coordinates it assumes the form
dσ 2 = γµν (x)dxµ dxν . (14.24)
Since the determinant |gµν | =
- 0, we can construct a contravari-
ant metric tensor with the aid of equations
gµσ g σν = δµν . (14.25)
220
With the aid of tensors gµν and g λσ it is possible to raise and
to lower indices
Aν = g νσ Aσ , Aν = gνσ Aσ . (14.26)
Geodesic lines in Riemannian space

Geodesic lines in Riemannian space play the same role as
straight lines in Euclidean space. They are called extremal
lines. For defining an extremal we shall take advantage of vari-
ational calculus. The essence of variational calculus consists
in generalization of the concepts of maximum and minimum.
The issue is not finding the extremum of a function, but find-
ing the extremum of a functional, i.e. finding such functions,
that make it an extremum. The distance between close points
in Riemannian space is determined by the interval ds. The
quantity ds is not a total differential. The interval between
points a and b is
#b #b -
S= ds = gµν (x)dxµ dxν . (14.27)
a a
The extremum is determined by the relation

#b #b
δ ds = δ(ds) = 0. (14.28)
a a
Thus, such functions gµν (x) are sought, that provide for the
functional (integral) achieving its extremum:
δ(ds2 ) = 2dsδ(ds) = δ(gµν (x)dxµ dxν ) =

∂gµν σ µ ν
= δx dx dx + 2gµν (x)dxµ δ(dxν ). (14.29)
∂xσ
221
We note that
δ(dxν ) = d(δxν ). (14.30)
On the basis of (14.29) and (14.30) we have
1 ∂gµν µ ν σ µ ν µ dxµ
δ(ds) = U dx δx +g µν U d(δx ), U = . (14.31)
2 ∂xσ ds
#b ( ν
)
1 ∂gµν µ ν σ µ d(δx )
δS = U U δx + g µν U ds = 0. (14.32)
a
2 ∂xσ ds
Since
d(δxν ) d d
gµν U µ = (gµν U µ δxν ) − δxν (gµν U µ ), (14.33)
ds ds ds
and at the integration limits δxν = 0, from (14.32) we obtain
#b (
1 ∂gµν µ ν dU µ
δS = U U − g µσ −
a
2 ∂xσ ds
∂gµσ
− λ U µ U λ ] dsδxσ = 0. (14.34)
∂x
We now represent the last term in (14.34) as
$ %
µ ∂gµσ 1 ∂gµσ ∂gλσ
U Uλ λ = + U µU λ. (14.35)
dx 2 ∂xλ ∂xµ
Substituting (14.35) into (14.34) we find
#b ( $ %
µ λ1 ∂gµσ ∂gλσ ∂gµλ
δS = U U + − +
a
2 ∂xλ ∂xµ ∂xσ
dU µ
)
+gµσ dsδxσ = 0. (14.36)
ds
222
Since the variation δxσ is arbitrary, the integral (14.36) turns
to zero, only if
dU µ 1
$ %
∂gµσ ∂gλσ ∂gµλ
gµσ + λ
+ µ
− σ
U µ U λ = 0. (14.37)
ds 2 ∂x ∂x ∂x
Multiplying (14.37) by g σα we obtain
dU α
+ Γαµλ U µ U λ = 0, (14.38)
ds
where the Christoffel symbols Γαµν are
1
Γαµλ = g ασ (∂λ gµσ + ∂µ gλσ − ∂σ gµλ ). (14.39)
2
The Christoffel symbols are not tensor quantities. Pre-
cisely equations (14.38) are the equations for a geodesic line.
There are four of them, but not all are independent, since the
following condition takes place:
gµν (x)U µ U ν = 1. (14.40)
By transformations of coordinates xµ it is possible to equate

the Christoffel symbols to zero along any not self-intersecting
chosen line [22].
Covariant differentiation
We now take an arbitrary covariant vector Aλ and form its
convolution with the vector U λ , and thus obtain the scalar
Aλ U λ , (14.41)
upon differentiating it with respect to ds we also have a scalar:

d dAλ λ dU ν
(Aλ U λ ) = U + Aν =
ds ds ds
∂Aλ σ λ dU ν
= U U + A ν . (14.42)
∂xσ ds
223
Substituting into the right-hand side expression (14.38) we
obtain
( )
d ∂Aλ
(Aλ U λ ) = σ
− Γνσλ Aν U σ U λ . (14.43)
ds ∂x
Since (14.43) is a scalar, and U σ is a vector, we hence have a

second-rank tensor
DAλ ∂Aλ
Aλ;σ = σ
= σ
− Γνσλ Aν . (14.44)
dx ∂x
Here and further a semicolon denotes covariant differenti-
ation. Thus, we have defined the covariant derivative of the
covariant vector Aλ . We shall now define the covariant deriva-
tive of the contravariant vector Aλ .
To this end we write the same scalar in the form
d µ ν ∂Aµ σ ν
(A U gµν ) = U U gµν +
ds ∂xσ
dU λ
+Aµ gµλ + Aµ U ν U σ ∂σ gµν . (14.45)
ds
Substituting into the right-hand side expression (14.38) we
obtain
∂Aµ
(
d µ ν
(A U gµν ) = U ν U σ gµν σ −
ds ∂x
)
µ
−A gµλ Γλσν µ
+ A ∂σ gµν . (14.46)
Taking into account expression (14.39) we have
∂Aµ
(
d µ ν
(A U gµν ) = gµν σ +
ds ∂x
)
1
+ (∂σ gµν + ∂µ gσν − ∂ν gσµ )A U ν U σ .
µ
(14.47)
2
224
Representing U ν in the form
U ν = Uλ g λν (14.48)
and substituting it into relation (14.47) we obtain
∂Aλ
( )
d µ ν
(A U gµν ) = σ
+ Γλσµ Aµ U σ Uλ . (14.49)
ds ∂x
Since this expression is a scalar, hence it follows that the con-

travariant derivative is a tensor,
DAλ ∂Aλ
Aλ;σ = = + Γλσµ Aµ . (14.50)
dxσ ∂xσ
Thus, we have defined the covariant derivative of the con-
travariant vector Aλ .
Applying formulae (14.44) and (14.50), it is also possible
to obtain covariant derivatives of a second-rank tensor:
∂Aµν
Aµν;σ = σ
− Γλσµ Aλν − Γλσν Aλµ , (14.51)
∂x
∂Aµν
Aµν
;σ = + Γµσλ Aνλ + Γνσλ Aµλ . (14.52)
∂xσ
∂Aνρ
Aνρ;σ = − Γλρσ Aνλ + Γνσλ Aλρ . (14.53)
∂xσ
Making use of expression (14.51) it is easy to show, that
gµν;σ ≡ 0,
i.e. the covariant derivative of a metric tensor is equal to zero.

The Riemann-Christoffel curvature tensor
In Riemannian space the operation of covariant differentiation
is noncommutative. Covariant differentiation of vector Aλ ,
225
first, with respect to the variable xµ and, then, with respect
to xν leads to the following expression:
∂Aλ;µ
Aλ;µν = − Γτνλ Aτ ;µ − Γτµν Aλ;τ , (14.54)
∂xν
but since
∂Aλ ∂Aτ
Aλ;µ = µ
− Γτλµ Aτ , Aτ ;µ = − Γσµτ Aσ ,
∂x ∂xµ (14.55)
∂Aλ σ
Aλ;τ = − Γλτ Aσ ,
∂xτ
upon substitution of these expressions into (14.54) we have
∂ 2 Aλ ∂Aτ ∂Aτ ∂Aλ

Aλ;µν = µ ν
− Γτλµ ν − Γτνλ µ − −Γτµν τ −
∂x ∂x ∂x ∂x ∂x
∂Γτλµ
−Aτ + Γτνλ Γσµτ Aσ + Γτµν Γσλτ Aσ . (14.56)
∂xν
We shall now calculate the quantity Aλ;νµ :
∂Aλ;ν
Aλ;νµ = − Γτµλ Aτ ;ν − Γτµν Aλ;τ , (14.57)
∂xµ
with account of the expression
∂Aλ ∂Aτ
Aλ;ν = ν
− Γτλν Aτ , Aτ ;ν = − Γστν Aσ ,
∂x ∂xν
(14.58)
∂Aλ
Aλ;τ = − Γσλτ Aσ ,
∂xτ
relation (14.57) assumes the form
∂ 2 Aλ ∂Aτ ∂Aτ ∂Aλ

Aλ;νµ = µ ν
− Γτλν µ − Γτµλ ν − Γτµν τ −
∂x ∂x ∂x ∂x ∂x
∂Γτλν
−Aτ + Γτµλ Γσντ Aσ + Γτµν Γσλτ Aσ . (14.59)
∂xµ
226
On the basis of (14.56) and (14.59), only the following terms
are retained in the difference:
∂Γσλν ∂Γσλµ
(
Aλ;µν − Aλ;νµ = Aσ − +
∂xµ ∂xν
)
+Γτνλ Γσµτ − Γτµλ Γσντ . (14.60)
σ
The quantity Rλµν is termed the Riemann curvature tensor
σ ∂Γσλν ∂Γσλµ
Rλµν = − + Γτνλ Γσµτ − Γτµλ Γσντ . (14.61)
∂xµ ∂xν
From this tensor it is possible, by convolution, to obtain a
second-rank tensor, the Ricci tensor:
σ ∂Γσλν ∂Γσλσ
Rλν = Rλσν = − + Γτνλ Γσστ − Γτσλ Γσντ . (14.62)
∂xσ ∂xν
We note that for an interval of the form (14.23) or (14.24) the
curvature tensor equals zero.
From expression (14.61) it is obvious that the curvature
tensor is antisymmetric with respect to the two last indices
µ, ν:
σ σ
Rλµν = −Rλνµ
It is possible to construct a curvature tensor with lower indices:
σ
Rρλµν = gρσ Rλµν .
It possesses the following symmetry properties:
Rρλµν = −Rλρµν = −Rρλνµ , Rρλµν = Rµνρλ .
We see that the curvature tensor is antisymmetric both with

respect to the first pair of indices and with respect to the
227
second. It is also symmetric with respect to permutation of
index pairs, without any change of their order.
In Riemannian space there exists a local coordinate sys-
tem, within which the first derivatives of the components of
the metric tensor gµν are equal to zero. Here, the Christoffel
symbols are, naturally, also equal to zero. Such coordinates
are called Riemann coordinates. They are convenient for find-
ing tensor identities, since if it has been established, that in
this coordinate system a certain tensor is zero, then, by virtue
of tensor transformations, it will also be zero in any coordinate
system.
The curvature tensor in a Riemann coordinate system is
σ
Rλµν = ∂µ Γσλν − ∂ν Γσλµ . (14.63)
The covariant derivative of the curvature tensor has the form

σ
Rλµν;ρ = ∂ρ ∂µ Γσλν − ∂ρ ∂ν Γσλµ . (14.64)
Cyclically transposing indices µ, ν, ρ and adding up the ob-

tained expressions we obtain the Bianchi identity
σ σ σ
Rλµν;ρ + Rλρµ;ν + Rλνρ;µ ≡ 0. (14.65)
Performing convolution of indices σ and ν we obtain

σ
− Rλµ;ρ + Rλρµ;σ + Rλρ;µ = 0. (14.66)
We multiply this expression by g λα :

α
−Rµ;ρ + (g λαRλρµ
σ α
);σ + Rρ;µ = 0.
We have, here, taken into account the previously established

property of metric coefficients consisting in that they can, in
228
case of covariant differentiation, be freely brought or taken out
from under the derivative sign.
Performing convolution of indices ρ and α we obtain
ρ
− Rµ;ρ + (g λρ Rλρµ
σ
);σ + ∂µ R ≡ 0, (14.67)
where
R = Rρρ = Rµν g µν
is the scalar curvature.
Let us consider under the derivative sign the second term
in identity (14.67):
g λρ Rλρµ
σ
= g λρ g νσ Rνλρµ = g νσ g λρ Rλνµρ = g νσ Rνµρ
ρ
= −Rµσ .
We have, here, applied the symmetry properties of the curva-

ture tensor and the definition of the tensor Rµν . Substituting
this expression into (14.67) we obtain
1 1
(Rµρ − δµρ R);ρ = ∇ρ (Rµρ − δµρ R) ≡ 0. (14.68)
2 2
We now introduce the notation
1
Gρµ = Rµρ − δµρ R. (14.69)
2
On the basis of (14.53), identity (14.68) can be written in the
expanded form
∂Gνρ
∇ν Gνρ = Gνρ;ν = ν
− Γλρν Gνλ + Γννλ Gλρ ≡ 0, (14.70)
∂x
taking into account that
1 ∂gµν
Γννλ = g µν λ (14.71)
2 ∂x
229
and differentiating the determinant g,
∂g ∂gµν
λ
= gg µν λ , (14.72)
∂x ∂x
we find, by comparison of (14.71) and (14.72), the following:
1 1 ∂g 1 √
Γννλ = · = √ ∂λ ( −g). (14.73)
2 g ∂xλ −g
√ √ √
∇ν ( −gGνρ ) = ∂ν ( −gGνρ ) − −gΓλρν Gνλ ≡ 0. (14.74)
Making use of expression (14.39), we find for the Christoffel

symbol
√ 1 ∂g λσ √
∂ν ( −gGνρ ) + · −gGλσ ≡ 0. (14.75)
2 ∂xρ
Such an identity was first obtained by D. Hilbert. It was neces-
sary for constructing the equations of general relativity theory.
In conclusion we shall show that the quantity determining
volume # -
!
v = −g ! dx0 ! dx1 ! dx2 ! dx3 ! (14.76)
is an invariant under arbitrary transformations of coordinates.
Under coordinate transformations we have
! ∂xλ ∂xσ
gµν (x! ) = gλσ (x) !µ · !ν .
∂x ∂x
We write this expression in the form
! ∂xλ ∂xσ
gµν (x! ) = · g λσ .
∂x!µ ∂x!ν
We shall, now, calculate the determinant g ! = det gµν
!
∂xλ ∂xσ
$ % $ %
g ! = det gσλ !µ det =
∂x ∂x!ν
∂xλ ∂xσ
$ % $ %
= det(gλσ ) det det .
∂x!µ ∂x!ν
230
Hence, we have
g ! = gJ 2. (14.77)
Here J is the transformation Jacobian,
∂(x0 .x1 , x2 , x3 )
J= . (14.78)
∂(x0! , x1! , x2! , x3! )
Thus, - √
−g ! = −gJ. (14.79)
√ ∂(x0 .x1 , x2 , x3 )
#
!
v = −g 0! 1! 2! 3!
dx0 ! dx1 ! dx2 ! dx3 ! =
∂(x , x , x , x )
#
√
= −gdx0 dx1 dx2 dx3 . (14.80)
But the right-hand side represents volume

#
√
v= −gdx0 dx1 dx2 dx3 . (14.81)
Thus, we have established the equality
v ! = v. (14.82)
Hence it follows, that the quantity

√
−gd4 x (14.83)
is also an invariant relative to arbitrary coordinate transfor-

mations.
Certain special features of Riemannian geometry should be
noted. In the general case, Riemannian space cannot be de-
scribed in a sole coordinate system. For its description an atlas
of maps is necessary. Precisely for this reason, the topology of
231
Riemannian space differs essentially from the topology of Eu-
clidean space. In the general case, no group of motion exists
in Riemannian space. In pseudo-Euclidean space, described
by the interval (14.23) or (14.24), there exists a ten-parameter
group of space motions.
The main characteristic of Riemannian geometry — the
σ
curvature tensor Rλµν — is a form-invariant quantity relative
to coordinate transformations. The tensor Rλν is also a form-
invariant quantity. Here, form-invariance is not understood
as one and the same functional dependence of the curvature
tensor upon the choice of coordinate system, but identity in
constructing the curvature tensor for a given expression gµν (x),
similarly to how expression
Aν (x)
is written in the same way in Galilean coordinates in differing

inertial reference systems for a given expression Aν . There
exists an essential difference between invariance and form-
invariance. for example, the operator γ µν (x)Dµ Dν (where
γ µν (x) is the metric tensor of Minkowski space) for arbitrary
coordinate transformations is an invariant, i.e. a scalar, but
it is not form-invariant. It will be form-invariant only in the
case of such coordinate transformations, under which the ten-
sor γ µν (x) remains form-invariant, i.e.
δγ µν (x) = 0.
The curvature tensor varies under gauge transformations

(3.16) according to the following rule:
δ- Rµναβ = −Rσναβ Dµ %σ − Rµσαβ Dν %σ −
−Rµνσβ Dα %σ − Rµνασ Dβ %σ − %σ Dσ Rµναβ .
232
This variation is due to arbitrary coordinate systems not being
physically equivalent.
Within the text of this book there are encountered, to-
gether with covariant derivatives in Riemannian space, ∇λ ,
covariant derivatives in Minkowski space, Dλ . The difference
consists in that in constructing the covariant derivatives Dλ
it is necessary to substitute into formulae (14.50 - 14.53) the
ν
Christoffel symbols of Minkowski space, γαβ , instead of the
Christoffel symbols of Riemannian space, Γναβ ,
In conclusion, we present the Weyl–Lorentz–Petrov theo-
rem [21]. Coincidence of the respective equations of isotropic
and timelike geodesic lines for two Riemannian spaces with
!
the metrics gµν (x) and gµν (x) and with the same signature –2
leads to their metric tensors only differing by a constant fac-
tor. From this theorem it follows that, if in one and the same
coordinate system x we have different metric tensors gµν (x)
!
and gµν (x), then, in identical conditions, different geodesic
lines, and, consequently, different physics, will correspond to
them. Precisely for this reason, the situation, that arises in
GRT with the appearance of a multiplicity of metrics within
one coordinate system, leads to ambiguity in the description
of gravitational effects.
ADDENDUM
On the gravitational force

The expression for the gravitational force is presented at page
66. We shall now derive this expression from the equation of a
geodesic in effective Riemannian space. The equation for the
geodesic line has the form
dpν dxν
+ Γναβ pα pβ = 0, pν = , ds2 = gµν dxµ dxν > 0. (1)
ds ds
In accordance with the definition of a covariant derivative in
Minkowski space we have
Dpν dpν ν
= + γαβ pα pβ . (2)
ds ds
Applying (1) and (2) we obtain
Dpν
= −Gναβ pα pβ . (3)
ds
Here
Gναβ = Γναβ − γαβ
ν
(4)
We shall write the left-hand side of relation (3) as
 
%2 d2 σ
ν ν
dxν
$
Dp dσ  DV
= + V ν 7 82  , V ν =
ds2
. (5)

ds ds dσ dσ

dσ
ds
Here V ν is the timelike velocity four-vector in Minkowski space,

that satisfies the condition
γµν V µ V ν = 1, dσ 2 > 0. (6)
Substituting (5) into (3) we obtain

d σ 2
DV ν ν α β ν ds2
= −Gαβ V V − V 7 82 (7)
dσ dσ
ds
234
From (6) we have
$ %2
dσ
= γαβ pα pβ . (8)
ds
Differentiating this expression with respect to ds we obtain
d2 σ
7
ds2
82 = −γλµ Gµαβ V λ V α V β . (9)
dσ
ds
Substituting this expression into (7) we find [4]

DV ν
= −Gµαβ V α V β (δµν − V ν Vµ ). (10)
dσ
Hence it is evident that the motion of a test body in Minkowski
space is due to the action of the force four-vector F ν :
F ν = −Gµαβ V α V β (δµν − V ν Vµ ), Vµ = γµσ V σ . (11)
One can readily verify that
F ν Vν = 0. (12)
By definition, the left-hand side of equation (10) is

DV ν dV ν ν
= + γαβ V αV β . (13)
dσ dσ
It must be especially noted that the motion of a test body
along a geodesic of effective Riemannian space can be rep-
resented as motion in Minkowski space due to the action of
the force F ν , only if the causality principle is satisfied. The
force of gravity and the Riemann curvature tensor, arising
from the gravitational equations (5.19) and (5.20), are cor-
related. Thus, if the curvature tensor is zero, then, by virtue
of equations (5.19) and (5.20), the gravitational force will also
235
be equal to zero. When the curvature tensor differs from zero
and Rµν -= 0, the force of gravity will also not be zero. And,
on the contrary, if the force of gravity F ν , arising from equa-
tions (5.19) and (5.20), differs from zero, then the Riemann
curvature is also not zero. Equating the gravitational force F ν
to zero results in the Riemann curvature tensor being equal to
zero.
Is the metric field of a non-inertial

reference system a special case of the
physical gravitational field?
From the causality conditions (6.10) and (6.11) it follows that,
if the vector Lν satisfies the condition
γµν Lµ Lν < 0, (1)
then the inequality

gµν Lµ Lν < 0. (2)
should also be fulfilled. We now form the convolution of equa-
tion (10.1) with the aid of the vector Lν defined by inequality
(1),
1
& '
2 µ ν
m γµν L L = 16π Tµν − gµν T −
2
µ ν 2 µ ν
−2Rµν L L + m gµν L L . (3)
Since we are only considering metric fields of Minkowski space,
equation (3) is simplified:
1
& '
2 µ ν
m γµν L L = 16π Tµν − gµν T + m2 gµν Lµ Lν . (4)
2
In the case of an ideal fluid, the energy-momentum tensor of
matter has the form
236
Tµν = (ρ + p)Uµ Uν − pgµν ,
(5)
dxµ
T = Tµν g µν = ρ − 3p, U ν = .
ds
Substituting (5) into (4) we obtain
m2 γµν Lµ Lν = 16π(ρ + p)(Uµ Lµ )2 −
m2
$ %
−8πgµν Lµ Lν ρ−p− . (6)
8π
From conditions (1) and (2) it follows that the right-hand side
of equation (6) is strictly positive, since
m2
ρ>p+ , (7)
8π
while the left-hand side of equation (6) is strictly negative.
Hence it follows that in the presence of matter no metric field
of Minkowski space satisfies the gravitational equations, and
therefore the metric fields arising in non-inertial reference sys-
tems of Minkowski space cannot be considered gravitational
fields. In the absence of matter, ρ = p = 0, equation (6) has
the sole solution
gµν (x) = γµν (x). (8)
On the covariant conservation law

The covariant conservation law of matter energy-momentum
tensor density Tµν in General Relativity (GRT) takes in Rie-
mannian space the following form
1 δLM
∇ν Tµν = ∂ν Tµν − T σλ ∂µ gσλ , T σλ = −2 . (1)
2 δgσλ
237
This equation is a straightforward consequence of Gilbert-
Einstein equations. Though the equation has a covariant form,
nevertheless the energy-momentum conservation law of matter
and gravitational field taken together has in GRT a noncovari-
ant appearance
∂ν (Tµν + τµν ) = 0. (2)
Just by this way the gravitational field pseudotensor τµν , which
is not a covariant quantity, arises in GRT. It is impossible
in principle to write conservation equations of the energy-
momentum of matter and gravitational field in the generally
covariant form. The idea that the gravitational energy cannot
be localized in GRT has arisen from this fact.
If we will not use Eq. (8.2) in the derivation of Eqs. (8.1),
then gravitational equations will take the following form
√ √
−γ(−J ελ + m2 φ̃ελ ) = 16π −g(T ελ + tελ
g ). (3)
Here tελ
g is the energy-momentum tensor density for the grav-
itational field.
√
16π −gtελ = −Dµ Dσ (φ̃ελ φ̃µσ − φ̃εµ φ̃λσ )+
+Dσ φ̃ελDµ φ̃µσ − Dµ φ̃εµ Dσ φ̃λσ +
+ 21 g ελgρτ Dµ φ̃αρ Dα φ̃µτ − gρτ g εµDµ φ̃αρ Dα φ̃λτ −
−gρτ7 g λν Dν φ̃αρ Dα φ̃ετ + αβ ερ
8 gρτ g Dα φ̃ Dβ φ̃ +
λτ
(4)
+ 21 gβρ gατ − 12 gβτ gαρ ×
7 8
× g εµg λν − 21 g ελg µν Dµ φ̃λρ Dν φ̃βτ −
7√ √ 8
−m2 −gg̃ ελ − −γ φ̃ελ + g̃ εα g̃ λβ γαβ − 21 g̃ ελ g̃ αβ γαβ .
J ελ = −Dµ Dν (γ µν g̃ ελ + γ ελ g̃ µν − γ εν g̃ µλ − γ εµ g̃ λν ). (5)
Let us mention that expression
Dσ (φ̃ελφ̃µσ − φ̃εµ φ̃λσ )

238
is antisymmetric under permutation of indicies λ µ, and so
the following identity takes place:
Dλ Dµ Dσ (φ̃ελφ̃µσ − φ̃εµ φ̃λσ ) = 0 .
It is also easy to get convinced, that the following equation is

valid
Dλ J σλ = 0 . (6)
Gravitational field equation (3) can be presented also in other
form (5.19)
√ 7 8
−g Rµν − 12 g µν R +
2
* 7 8 C (7)
+ m2 g̃ µν + g̃ µα g νβ − 12 g̃ µν g αβ γαβ = 8πT µν .
From Eqs. (7) it follows

√ & 1
'
m2 −g g µα g νβ − g µν g αβ ∇µ γαβ = 16π∇µ T µν . (8)
2
By taking into account relation
∇µ γαβ = −Gσµα γσβ − Gσµβ γσα , (9)
we find
7 8
g µα g νβ − 12 g µν g αβ ∇µ γαβ =
(10)
= −g µα g νβ Gσµα γσβ − g µα g νβ Gσµβ γσα + g µν g αβ Gσµα γσβ .
It is easy to see that the following identity takes place:
−g µα g νβ Gσµβ γσα + g µν g αβ Gσµα γσβ . (11)
Therefore we have
1
& '
µα νβ
g g − g µν g αβ ∇µ γαβ = −g µα g νβ Gσµα γσβ . (12)
2
239
By substituting expression (4.5) instead of Gσµα we get
7 8
g µα g νβ − 12 g µν g αβ ∇µ γαβ =
7 8 (13)
= γµλ g µν Dσ g σλ + Gσασ g αλ .
After taking into account Eq. (13), Eq. (8) takes the following
form
√ 7 8
16π∇µ T µν = m2 −g γµλ g µν Dσ g σλ + Gσασ g αλ . (14)
Applying equation
√ 7 8
−g Dσ g σλ + Gσασ g αλ = Dσ φ̃σλ , (15)
we find
m2 γνλ Dσ φ̃σλ = 16π∇µ Tνµ . (16)
According to Eq. (3) we have
5
2 σλ σλ g 7 σλ 8
m φ̃ =J + 16π T + tσλ
g . (17)
γ
Substituting Eq. (17) into Eq. (16) we get

(5 )
g 7 σλ 8
Dσ T + tσλ
g = γ νλ ∇µ Tνµ . (18)
γ
When matter equations of motion are valid we have

δLM
= 0. (19)
δφA
According to strong identity (.16) the following equality is

valid
∇µ Tνµ = 0 , (20)
240
and therefore, according to Eq. (18), the covariant conser-
vation law for energy-momentum of matter and gravitational
field taken together is as follows
*√ 7 8C
Dσ −g T σλ + tσλ
g = 0. (21)
So, if in GRT covariant law (1) leads to a noncovariant

conservation law for energy-momentum of matter and gravita-
tional field (2), and also to arising of a noncovariant quantity –
pseudotensor τµν of the gravitational field, then in RTG covari-
ant law (1) together with gravitational equations written in a
form (3) or (7) exactly leads to the covariant conservation law
for energy-momentum of matter and gravitational field writ-
ten in a form (21). In Eq. (21) the gravitational component
√
−g tσλ
g enters in additive form under the Minkowski space
covariant derivative symbol, whereas the gravitational compo-
nent disappears from Eq. (20), it is used for generating the
effective Riemannian space and so only the energy-momentum
tensor density of matter in Riemannian space stays under the
covariant derivative symbol. The presence of covariant conser-
vation laws (21) is just the point to see that the gravitational
energy as also all other forms of energy is localizable.
H.Poincare On the dynamics of the

electron
(5 June 1905)36. (The comments are italicized and indicated
by an asterisk).
36
Poincare H. Sur la dynamique de l’electron // Comptes rendus
hebdomadaires des seances de l’Akademie des sciences. – Paris, 1905. –
V.140. – P.1504–1508.
241
It seems at first sight that the aberration of light and the re-
lated optical and electrical phenomena would provide us with
a means of determining the absolute motion of the Earth, or
rather its motion not with respect to the other stars, but with
respect to the ether. Actually, this is not so: experiments
in which only terms of the first order in the aberration were
taken into account first yielded negative results, which was
soon given an explanation; but Michelson also, who proposed
an experiment in which terms depending on the square aberra-
tion were noticeable, also met with no luck. The impossibility
to disclose experimentally the absolute motion of the Earth
seems to be a general law of Nature.
* “Experiment has provided numerous facts justifying the
following generalization: absolute motion of matter, or, to be
more precise, the relative motion of weighable matter and ether
cannot be disclosed. All that can be done is to reveal the mo-
tion of weighable matter with respect to weighable matter” 37 .
The above words written by Poincare ten years earlier quite
clearly demonstrate that his vision of a general law determin-
ing the impossibility of absolute motion of matter had been
maturing since long ago.
In development of his idea on the total impossibility of
defining absolute motion in relation to the new hypothesis, put
forward by Lorentz and according to which all bodies should
experience a decrease in length by 1/2 · 10−9 in the direction of
motion of the Earth, Poincare wrote:
“Such a strange property seems to be a real coup de pouce
presented by Nature itself, for avoiding the disclosure of ab-
37
Poincare H. On Larmor’s theory // The relativity principle: Col-
lection of works on special relativity theory. — Moscow, 1973. — P.7.
242
solute motion with the aid of optical phenomena. I can’t be
satisfied and I must here voice my opinion: I consider quite
probable that optical phenomena depend only on the relative
motion of the material bodies present, of the sources of light
or optical instruments, and this dependence is not accurate
up to orders of magnitude of the square or cubic aberration,
but rigorous. This principle will be confirmed with increasing
precision, as measurements become more and more accurate.
Will a new coup de pouce or a new hypothesis be nec-
essary for each approximation? Clearly this is not so: a well
formulated theory should permit proving a principle at once
with all rigour. The theory of Lorentz does not permit this yet.
But, of all theories proposed it is the one nearest to achieving
this goal” 40 .
In a report to the Congress of art and science held in Saint
Louis in 1904 Poincare among the main principles of theoret-
ical physics formulates the relativity principle, in accordance
with which, in the words of Poincare, “the laws governing phys-
ical phenomena should be the same for a motionless observer
and for an observer experiencing uniform motion, so there is
no way and cannot be any way of determining whether one
experiences such motion or not”41 .
An explanation has been proposed by Lorentz, who has
introduced the hypothesis of a contraction experienced by all
bodies in the direction of the motion of the Earth; this con-
traction should account for Michelson’s experiment and for all
40
Poincare H. Electricite et optique: La lumiere et les theories elec-
trodynamiques. — 2 ed., rev. et complettee par Jules Blondin, Eugene
Neculcea. Paris: Gauthier–Villars, 1901.
41
Poincare H. L’etat et l’avenir de la Physique mathematique // Bul-
letin des Sciences Mathematiques. — Janvier 1904 — V.28. Ser.2 —
P.302–324; The Monist. — 1905. V.XV, N1.
243
other relevant experiments performed to date. It would, how-
ever, leave place for other even more subtle experiments, more
simple to be contemplated than to be implemented, aimed at
revealing absolute motion of the Earth. But considering the
impossibility of such a claim to be highly probable one may
foresee that these experiments, if ever they will be performed,
to once again provide a negative result. Lorentz has attempted
to complement and alter the hypothesis so as to establish a
correspondence between it and the postulate of total impos-
sibility of determining absolute motion. He has succeeded in
doing so in his article entitled Electromagnetic phenomena in
a system moving with any velocity smaller than that of light
(Proceedings de l’Academie d’Amsterdam, 27 May 1904).
The importance of this issue has induced me to consider
it once again; the results I have obtained are in agreeement
with those obtained by Lorentz in what concerns all the main
points; I have only attempted to modify them somewhat and
to complement them with some details.
The essential idea of Lorentz consists in that the equations
of the electromagnetic field will not be altered by a certain
transformation (which I shall further term the Lorentz trans-
formation) of the following form:
x! = γl(x − βt), y ! = ly, z ! = lz, t! = γl(t − βx), (1)
where x, y, z are the coordinates and t is the time before the

transformation; and x! , y !, z ! and t! are the same after the
transformation. The quantity β is a constant determined by
the transformation
1
γ=√ ,
1 − β2
while l is a certain function of β:
244
* Poincare writes: “The idea of Lorentz”, but Lorentz never
wrote such words before Poincare. Here, Poincare formulated
his own fundamental idea, but attributed it fully to Lorentz.
Probably more than any other person, did he always highly
esteem and note each person, who gave his thought an impetus
and presented him with the happiness of creativity. He was
totally alien to issues of his own personal priority.
From formulae (1) it is immediately seen that the condition
x = βt corresponds to the origin (x! = y ! = z ! = 0) of the new
reference system. In other words, the new origin is shifted in
the reference system x, y, z with a speed β along the x-axis.
Thus, the Lorentz tranformations relate the variables
(x, y, z, t) referred to one reference system and the variables
(x! , y !, z ! , t! ) in another system moving uniformly in a straight
line along the x-axis with the velocity β relative to the first
system.
Proof of the statement that the equations of the electro-
magnetic field do not alter under the Lorentz transformations
signifies that electromagnetic phenomena are described in both
reference systems by identical equations and that, consequently,
no electromagnetic processes can be utilized to distinguish be-
tween the (x, y, z, t) reference system and the (x! , y !, z ! , t! ) ref-
erence system moving uniformly in a straight line with respect
to the first system.
We see that invariance of the equations of the electromag-
netic field under transformations of the Lorentz group results
in the relativity principle being fulfilled in electromagnetic phe-
nomena. In other words, the relativity principle for electro-
magnetic phenomena follows from the Maxwell–Lorentz equa-
tions in the form of a rigorous mathematical truth.
245
In this transformation the x-axis plays a particular role,
but it is clearly possible to construct such a transformation,
in which this role will be assumed by a certain straight line
passing through the origin. The set of all such transformations
together with all spatial rotations should form a group; but for
this to take place it is necessary that l = 1; hence one is led
to assume l = 1, which is precisely the consequence obtained
by Lorentz in another way.
* It must be underlined that, by having established the group
nature of the set of all purely spatial transformations together
with the Lorentz transformations, that leave the equations of
electrodynamics invariant, Poincare thus discovered the exis-
tence in physics of an essentially new type of symmetry re-
lated to the group of linear space-time transformations, which
he called the Lorentz group.
Supplemented with transformations of space coordinate and
time translations, the Lorentz group forms a maximum group
of space-time transformations, under which all equations of
motion for particles and fields remain invariant and which now
is called the Poincare group, the name given to it subsequently
by E.Wigner. Richard Feynman wrote about this fact as fol-
lows: “Precisely Poincare proposed to find out what one can
do with equations without altering their form. He was the per-
son who had the idea to examine the symmetry properties of
physical laws”.
Let ρ be the charge density of the electron, and vx , vy , and
vz the components of the electron velocity before the trans-
formation; then, after applying the transformation one has for
246
these same quantities ρ! , v x ! , v y ! , and v z ! the following:
ρ! = γl−3 ρ(1 − βvx ), ρ! v x ! = γl−3 ρ(vx − β),
(2)
ρ! v y ! = l−3 ρvy , ρ! v z ! = l−3 ρvz .
These formulas differ somewhat from the ones found by
Lorentz.
Now, let f0 and f0! be the three-force components before and
after application of the transformation (the force is referred to
unit volume); then
0v ); f ! = l−5 fy ; f ! = l−5 fz .
fx! = γl−5 (fx − β f0 (3)
y z
These formulas also differ somewhat from the ones pro-

posed by Lorentz; the additional term in f0 · 0v reminds the
result earlier obtained by Lienard.
If we now denote by F0 and F0 ! the force components referred
to the electron mass unit, instead of unit volume, we obtain
ρ ρ ρ
Fx! = γl−5 ! (Fx − β F0 · 0v ); Fy! = ! l−5 Fy ; Fy! = ! l−5 Fx . (4)
ρ ρ ρ
* Formulae (2), (3), and (4) comprise the relativistic trans-
formation laws, first established by Poincare, for the charge
density and velocity of motion of an electron, referred both to
unit charge and unit volume.
Thus, Poincare, as compared with Lorentz, made the deci-
sive step in this work and laid down the foundations of rela-
tivity theory.
It is notable that, while developing totally new ideas in ar-
ticles on the dynamics of the electron and correcting and com-
plementing Lorentz, Poincare is careful to paid maximum trib-
ute to Lorentz as the discoverer and leaves it to others to judge
about his own personal contribution to the creation of relativity
theory.
247
The testimony of Lorentz himself is extremely important in
this respect, since it permits filling in the gap, sometimes left
by certain authors writing about the history of the creation of
relativity theory, and to do justice to Poincare as the creator
of relativistic mechanics and special relativity.
Thus, in discussing the relativistic transformation formu-
lae for the velocities, charge densities, and current of the elec-
tron, Lorentz wrote:46 “Formulae (4) and (7) are absent in
my article published in 1904, since I didn’t even think of a
direct path leading to them, because I thought an essential dif-
ference existed between the systems x, y, z, t and x! , y ! , z ! , t! . In
one of them – such was my reasoning – coordinate axes were
used that had a fixed position in ether and what could be called
true time; in the other system, on the contrary, one dealt with
simply auxiliary quantities introduced only with the aid of a
mathematical trick. Thus, for instance, the variable t! could
not be considered to be time in the same sense as the variable t.
Given such reasoning, I had no intention of describing phe-
nomena in the system x! , y !, z ! , t! in precisely the same manner
as in the system x, y, z, t...” And further, in the same work:
“... I was unable to achieve total invariance of the equations;
my formulae remained cumbersome owing to additional terms,
that should have disappeared. These terms were too small to
exert noticeable influence on the phenomena, and this sup-
plied me with an explanation for their being independent of
the Earth’s motion revealed by obs and universal truth.
46
Lorentz H.A. Two articles by Henri Poincare on mathematical
physics // The relativity principle: Collection of works on special rel-
ativity theory. – Moscow, 1973. – P.189–196.
Formulae (4) and (7) dealt with by Lorentz are the transformation for-
mulae for the electron velocities and charge densities, respectively.
248
Contrariwise, Poincare achieved total invariance of the equa-
tions of electrodynamics and formulated the ¡¡relativity pos-
tulate¿¿ — a term introduced by him. Indeed, adopting the
point of view, that I had failed to take into account, he derived
formulae (4) and (7). We should add, that in correcting the
defects of my work he never reproached me for them”.
Lorentz also arrived at the necessity of assuming a moving
electron to have the shape of a compressed ellipsoid; the same
hypothesis was made by Langevin, but while Lorentz assumed
the two axes of the ellipsoid to be constant, in agreement with
his hypothesis that l = 1, Langevin assumed, contrariwise, the
volume of the ellipsoid to be constant. Both authors showed
the two hypotheses to be in the same good agreement with
the experiments performed by Kaufmann, as the initial hy-
pothesis of Abraham (the spherical electron). The advantage
of Langevin’s hypothesis consists in its being sufficient, i.e. it
suffices to consider the electron to be deformable and incom-
pressible for explaining why it assumes an ellipsoidal shape in
motion. But I can show, without contradicting Lorentz, that
this hypothesis cannot be consistent with the impossibility of
revealing absolute motion. As I have already said, this occurs
because l = 1 is the only hypothesis for which the Lorentz
transformations form a group.
But in the Lorentz hypothesis, also, the agreement between
the formulas does not occur just by itself; it is obtained to-
gether with a possible explanation of the compression of the
electron under the assumption that the deformed and com-
pressed electron is subject to constant external pres-
sure, the work done by which is proportional to the
variation of volume of this electron.
249
Applying the principle of least action, I can demonstrate
the compensation under these conditions to be complete, if
inertia is assumed to be of a totally electromagnetic origin, as
generally acknowledged after Kaufmann’s experiments, and if
all forces are of an electromagnetic origin, with the exception
of the constant pressure of which I just spoke and which acts on
the electron. Thus, it is possible to explain the impossibility of
revealing the absolute motion of the Earth and the contraction
of all bodies in the direction of the Earth’s motion.
But this is not all. In the quoted work Lorentz considers
it necessary to complement his hypothesis with the assump-
tion that in the case of uniform motion all forces, of whatever
origin, behave exactly like electromagnetic forces, and that,
consequently, the influence of the Lorentz transformation on
the force components is determined by equations (4).
* Here, Poincare in development of the assumption expressed
by Lorentz extends the Lorentz transformations to all forces,
including, for instance, gravitational forces.
He was the first to point out that the relativity postulate
requires such a modification of the laws of gravity, according to
which the propagation of forces of gravity is not instantaneous,
but proceeds with the speed of light.
It has turned out necessary to consider more carefully this
hypothesis and, in particular, to clarify which changes it com-
pels us to introduce into the laws of gravity. This is just what
I attempted to determine: I was first induced to assume the
propagation of gravity forces to proceed with the speed of light,
and not instantaneously. This seems to contradict the result
obtained by Laplace who claims that although this propaga-
tion may not be instantaneous, it is at least more rapid than
250
the propagation of light. However, the issue actually raised
by Laplace differs significantly from the issue dealt with here
by us. According to Laplace, a finite propagation velocity was
the sole alteration, introduced by him to Newton’s law. Here,
also, a similar change is accompanied by many others; hence,
partial compensation between them is possible, and it actually
does take place.
Consequently, if we speak about the position or velocity of
a body exerting attraction, we shall bear in mind its position
or velocity at the moment, when the gravitational wave
departs from this body; if we speak about the position or
velocity of a body being attracted, we shall intend its position
or velocity at the moment, when this body being attracted is
overcome by the gravitational wave emitted by another body:
the first moment clearly precedes the second.
Hence, if x, y, z are the projections onto three axes of the
vector 0r connecting the two positions and if 0v = (vx , vy , vz )
are the velocity components of the body attracted and v01 =
(v1x , v1y , v1z ) are the velocity components of the attracting
body, then the 1z three components of the attraction (which
I may also call F0 ) will be functions of 0r , 0v, v01 . The question
is whether these functions can be defined in such a way that
they behave under the Lorentz transformation in accordance
with equations (4) and that the conventional law of gravity be
valid in all cases of the velocities 0v , v01 being sufficiently small
to allow neglecting their square values as compared with the
square speed of light?
The answer to this question must be affirmative. It has
been revealed that the attraction, taking into account the cor-
rection, consists of two forces, one of which is parallel to the
251
components of the vector 0r , and the other to the components
of the velocity v01 .
The disagreement with the conventional law of gravity, as
I just pointed out, is of the order of v 2 ; if, on the other hand,
one assumes, as Laplace did, the propagation velocity to be
equal to the speed of light, this divergence will be of the order
of v, i.e. 10000 times greater. Consequently, at first sight, it
does not seem absurd to assume astronomical observations to
be insufficiently precise for revealing the smallest imaginable
divergence. Only a profound investigation can resolve this
issue.
* Poincare thus introduces the physical concept of gravita-
tional waves, the exchange of which generates gravitational
forces, and supplies an estimation of the contribution of rela-
tivistic corrections to Newton’s law of gravity.
For example, he shows that the terms of first order in v/c
cancel out exactly and so the relativistic corrections to New-
ton’s law are quantities of the order of (v/c)2 .
These results remove the difficulty noted previously by Laplace
and permit making the conclusion that the hypothesis equating
the speeds of light and of gravitational influence is not in con-
tradiction with observational data.
Thus, in this first work Poincare already gave a general
and precise formulation of the main points of relativity theory.
It is here that such concepts as the following first appeared:
the Lorentz group, invariance of the equations of the electro-
magnetic field with respect to the Lorentz transformations, the
transformation laws for charge and current, the addition for-
mulae of velocities, the transformation laws of force. Here,
also, Poincare extends the transformation laws to all the forces
of Nature, whatever their origin might be.
253
Bibliography
[1] Weinberg S. Gravity and cosmology. Moscow: Mir, 1975.
[2] Vlasov A.A., Logunov A.A. Theor. and Math. Phys.,
1989, vol.78, numb.3, pp.323–329.
[3] Vlasov A.A., Logunov A.A. Mestvirishvili M.A., Theor.
and Math. Phys. 1984, vol.61, numb.3, pp.323–326.
[4] Landau L.D., Lifshitz E.M., Field theory. Moscow:
Nauka, 1973.
[5] Logunov A.A. Theor. and Math. Phys. 1989, vol.80,
numb.2, pp.165–172.
numb.2, pp.191–206.
[7] Logunov A.A. Lectures in relativity and gravity theory.
Modern analysis of the problem. Moscow: Nauka, 1987.
[8] Logunov A.A., Loskutov Yu.M. Doklady AN SSSR, 1989,
vol.305, numb.4, pp.848–851.
[9] Logunov A.A. Mestvirishvili M.A., Theor. and Math.
Phys. 1984, vol.61, numb.3, pp.327–345.
[10] Logunov A.A. Mestvirishvili M.A., Relativistic theory of
gravity, Moscow: Nauka, 1989.
Phys. 1991, vol.86, numb.1, pp.3–15.
Phys. 1997, vol.110, numb.1, pp.5–24.
Phys. 1999, vol.121, numb.1, pp.4–24.
254
[14] Loskutov Yu.M. Theor. and Math. Phys. 1990, vol.82,
numb.2, pp.304–312.
[15] Loskutov Yu.M. Vest. MGU, Series 3, Fizika, astronomia,
1991, vol.32, numb.4, p.49.
[16] Loskutov Yu.M. Theor. and Math. Phys. 1996, vol.107,
numb.2, pp.329–343.
[17] Mandel’stam L.I. Lectures in optics, relativity theory and
quantum mechanics, Moscow: Nauka, 1972.
[18] Mach E. Mechanics: Historical and critical essay on its
evolution// In: A.Einstein and the theory of gravity.
Moscow: Mir, 1979.
[19] Moeller K. Relativity theory. Moscow: Atomizdat, 1975.
[20] Pauli W. Relativity theory. Moscow: Gostekhizdat, 1947.
[21] Petrov A.Z. New methods in general relativity theory.
Moscow: Nauka, 1966.
[22] Rashevsky P.K. Riemannian geometry and tensor analy-
sis. Moscow: Gostekhizdat, 1953.
[23] Synge J.L. Relativity: the general theory. Moscow: Izda-
tel’stvo inostr.lit., 1963.
[24] Fock V.A. JETP, 1939, vol.9, numb.4, p.375.
[25] Fock V.A. Theory of space, time and gravity. Moscow:
Gostekhizdat, 1961.
[26] Hawking S., Ellis G. Large-scale structure of space-time.
Moscow: Mir, 1977.
[27] Einstein A. Collection of scientific works, Moscow:
Nauka, 1965, Vol.1, p.22.
Nauka, 1965, vol.1, p.21.
255
Nauka, 1965, vol.1, pp.21, 28, 29, 32.
[30] Bohm D. The special theory of relativity. N.Y.:Benjamin,
1965.
[31] Gerstein S.S., Logunov A.A. Mestvirishvili M.A.// Phys.
Atomic Nuclei. 1998. Vol.61, numb.8, pp.1420–1429.
[32] Hawking S.W., Penrose R. Proc. Roy. Soc., Ser.A. 1970,
vol.314, p. 529.
[33] Kolb E.W., Turner M.S., The Early Universe. L.:Addison-
Wesley, 1990.
numb.3, pp.1184-1187.
[35] Logunov A.A. Relativistic theory of gravity and the Mach
principle. Dubna, 1997.
[36] Logunov A.A. Phys. Particles and Nuclei. 1998. Vol.29,
numb. 1
[37] Logunov A.A. Relativistic theory of gravity. N.Y.: Nova
Science Publ., 1998. (Horizons in World Phys.; Vol.215).
[38] Logunov A.A. Mestvirishvili M.A., The relativistic theory
of gravitation, Moscow: Mir, 1989.
[39] Loskutov Yu.M., Proc. of the VI Marcel Grossman meet-
ing on gen.relativ. 1991. Pt B. pp.1658–1660.
[40] Poincaré H. Bull. Sci. Math. Ser.2 1904. Vol.28. pp.302–
328.
[41] Rosen N. Phys.Rev. 1940. Vol.57, p.147.
[42] Pugh G.E. WSEG Res. Mem. US Dep. of Defense. 1959.
N 11.
256
[43] Shapiro I.I. Centenario di Einstein: Astrofisica e cosmolo-
gia gravitazione quanti e relativita. Firenze: Giuni Bar-
bera, 1979.
[44] Schiff L.I. Proc. Nat. Acad. Sci. US. 1960. Vol.46. p.871;
Phys. Lett. 1960. Vol.4. p.215.
[45] Weinberg S. Gravitation and Cosmology. N.Y. etc., 1972.

Logunov. The Theory of Gravity PDF

Uploaded by

Copyright:

Available Formats

Logunov. The Theory of Gravity PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Logunov. The Theory of Gravity PDF

Uploaded by

Copyright:

Available Formats

Contents

dσ 2 = (dX 0 )2 − (dX 1 )2 − (dX 2 )2 − (dX 3 )2 . (α)

dσ 2 = γµν (x)dxµ dxν , where

is not a complete differential, so the synchronization of clocks

ds2 = gik dxi dxk ” 7 .

A.Einstein wrote something similar in 1930: “In special rel-

and he further pointed out: ¡¡The motion of a material point in

Dµ tµν = ∂µ tµν + γαβ

Here tµν is the total conserved density of the energy-momentum

The introduction of densities is due to an invariant volume el-

while an invariant volume element in Riemannian space is

Therefore, the principle of least action assumes the form

Owing to gravity being universal, it would be nat-

Ãν = (φ̃, Ã) .

In the absence of gravity, Maxwell’s equations of electrody-

γ αβ Dα Dβ Ãν + µ2 Ãν = 4πj ν ,

Here, for generalization we have introduced the parameter µ,

γ αβ Dα Dβ φ̃µν + m2 φ̃µν = λtµν , (2.2)

Here λ is a certain constant which, in accordance with the

The interaction between the gravitational field and matter is

L = Lg (γµν , φ̃µν ) + LM (γµν , φ̃µν , φA ) ,

here Lg is the density of the Lagrangian of the gravitational

Equations (2.4) differ from equations (2.2), first of all, in that

Thus arises the effective Riemannian space with the

It must be stressed that condition (2.6) permits substitut-

γ00 > 0, dl2 = sik dxi dxk > 0; i, k = 1, 2, 3,

it is easy to find the group of transformations, under which

where the expressions

δξ g̃ µν (x) = g̃ µα Dα ξ ν (x) + g̃ ναDα ξ µ (x) − Dα (ξ α g̃ µν ),

are Lie variations.

[δξ1 , δξ2 ](·) = δξ3 (·) (3.6)

and the Jacobi identity

The variation of action under the coordinate transforma-

The first integral in (3.11) can be written in the form

Owing to the integration volume Ω being arbitrary, we have

δ LM (x) = −∂α (ξ α (x)LM (x)), (3.13)

where the Lie variation δLM is

δL(g̃ µν (x)) = −∂α (ξ α (x) L(g̃ µν (x))), (3.13a)

where the Lie variation δL is

The Lie variations (3.5) were established within the con-

g̃ µν (x) → g̃ µν (x) + δg̃ µν (x),

with the variations

δε g̃ µν (x) = g̃ µα Dα εν (x) + g̃ να Dα εµ (x) − Dα (εα g̃ µν ),

Like in the preceding case, we have

εν3 = εµ1 Dµ εν2 − εµ2 Dµ εν1 = εµ1 ∂µ εν2 − εµ2 ∂µ εν1 .

The gauge group arose from the geometrized structure of

δε g̃ µν (x) = δε φ̃µν (x).

From (3.16) the transformation for the field follows

δε φ̃µν (x) = g̃ µα Dα εν (x) + g̃ να Dα εµ (x) − Dα (εα g̃ µν ),

δξ φ̃µν (x) = φ̃µα Dα ξ ν (x) + φ̃να Dα ξ µ (x) − Dα (ξ α φ̃µν ).

Under the gauge transformations (3.16) the equations of mo-

R̃ = −g̃ µν (Γλµν Γσλσ − Γλµσ Γσνλ ) − ∂ν (g̃ µν Γσµσ − g̃ µσ Γνµσ ). (4.4)

Since the Christoffel symbols are not tensor quantities, no

Note that under arbitrary coordinate transformations each

varies only by the divergence under arbitrary gauge transfor-

we exclude from the preceding expression terms containing

Thus, we see that the requirement for the density of the

δ- Rµν = −Rµσ Dν %σ − Rνσ Dµ %σ − %σ Dσ Rµν ,

To retain the concept of a field in Minkowski space and