Vous êtes sur la page 1sur 30

The Two Schwarzschild Solutions: A Critical

Assessment
R.E. Salvino
# 604
1 South Shamian Street
Guangzhou, China 510133

R.D. Puff
Department of Physics, Box 351560
University Of Washington
Seattle WA 98195

Revised: 29 Sep 2013

Abstract
We present a pedagogically sound derivation of the most general solution of the
time-independent, spherically-symmetric gravitational field equations. We use that
solution, the Combridge-Janne solution, as a basis for evaluating the original and textbook Schwarzschild solutions. We demonstrate that both versions of the Schwarzschild
solution are valid, are distinct and not equivalent to each other, but are related by
means of a one-parameter family of solutions. We explicitly show that the original
solution is the appropriate solution for a point mass source while the textbook solution is the appropriate solution for a wormhole source. In addition, the textbook
solution necessarily has a time-dependent aspect while the original solution is truly
time-independent. A number of issues surrounding these two solutions are clarified
and resolved.
Keywords: Combridge-Janne solution, generalized Schwarzschild solution, Schwarzschild
solution, original Schwarzschild solution, Hilbert solution, textbook Schwarzschild solution

Introduction

It is now known that the Schwarzschild solution, the first and best known exact solution
to the general relativistic field equations, actually comes in two versions. The first version,
which we will temporarily call the textbook solution [1,2] or the Hilbert solution [3,4], is by

Current address: 9 Thomson Lane, 15-06 Sky@Eleven, Singapore 297726.

far the most familiar, has the well-known coordinate singularity and event horizon at the
Schwarzschild radius, was the origin for the concept of a singularity in spacetime, and has
the preferred interpretation as the gravitational field of a wormhole in spacetime [1, 2, 5].
The second version, which we will call the original Schwarzschild solution [68], is regular
everywhere, has a physical singularity and event horizon at the single point r = 0, and has
the simple interpretation as the gravitational field for a point mass located at r = 0 [69]
The basic distinction between these two Schwarzschild solutions is traced to a single
metric function, g22 . The textbook solution develops from three general relativistic field
equations in two unknown metric functions, g00 and g11 . The g22 function, in effect, is chosen to be g22 = r2 by means of an argument based on a suitable choice of coordinates
and is not determined by a field equation [1, 2]. It also satisfies the asymptotic boundary
condition that the gravitational field be described by the Newtonian theory far from the
source. The original Schwarzschild solution develops from four equations in three unknown
metric functions, g00 , g11 , and g22 . The four equations consist of the three general rela
tivistic field equations plus the subsidiary condition g = 1 where g is the determinant
of the metric [68]. While this additional condition was originally part of the structure of
the theory of general relativity and later abandoned, Einstein continued to promote this
additional equation as defining a particularly convenient choice of coordinates that may
simplify resulting equations [10]. Consequently, we view Schwarzschilds approach as uti
lizing a choice of coordinates such that g = 1 and, having completed the solution in
those coordinates, then expressing the solution in the original contravariant coordinates.
In addition, Schwarzschilds solution was designed to satisfy two boundary conditions. The
first is the same asymptotic condition that the textbook solution satisifies, that the field
be described by Newtonian theory far from the source. The second is the condition that
the metric functions be finite and continuous everywhere except at r = 0, the presumed
site of the point mass source object. Schwarzschilds original solution then states that
g22 = Rs2 (r) = (r3 + 3 )2/3 where = 2mG and mG = Gm/c2 is the geometric mass of
the object [68].
The difference between the number of boundary conditions imposed to uniquely determine the solutions is important. The field equations consist of a system of differential
equations and the highest order of those equations is two. Consequently, we expect that
two boundary conditions are needed to uniquely determine the solution. As we pointed out
in the above discussion, Schwarzschild did originally impose two boundary conditions while
the textbook presentation imposes only one. However, as we show below in Section 7, the
textbook condition that g22 = r2 everywhere serves as the necessary second boundary
condition for the textbook solution. It is not recognized as a boundary condition because
it is imposed before the field equations are even established and its role is obscured by the
claim that the condition is the result of a choice of coordinates.
In both versions of the solution, only two field equations are necessary in the determination of the solution, but the third field equation is satisfied by that solution. Textbook
presentations often state that a third field equation is not needed and it is a common
2

exercise to verify that the textbook solution does indeed satisfy the third field equation.
Schwarzschild had pointed out in his original paper that only two of the three field equations, with the addition of the auxiliary condition, were needed, but he verified that the
solution did satisfy the third equation. The fact that the third field equation is not needed
in either case and yet is satisfied by both solutions is never explored and is simply accepted
as a pleasant fact. It is not, however, just a matter of luck. In 1916, within the context of a
perturbative treatment of the time-independent inhomogeneous and spherically symmetric
gravitation problem, de Sitter showed that there are only two independent field equations,
not three [11]. This is an exact and rigorous statement in the limit of vacuum field conditions. This means that the third field equation is necessarily satisifed by the solution since
it is a consequence of the other two equations. Furthermore, it should be expected that the
vacuum field equations will produce a solution for two of the metric functions in terms of
the third metric function while the third metric function remains undetermined by the field
equations. This, in fact, was explicitly demonstrated independently by Combridge [12] and
Janne [13] in 1923.
The solution provided by Combridge and Janne provides the basis for understanding the
the relationship between the two Schwarzschild solutions. Only two equations are needed
for the solution because the third field equation is not independent of the other two. Thus,

imposing the condition g = 1 provides Schwarzschild with the basis for determining
the third metric function; stating that g22 = r2 everywhere appears to provide the third
metric function simply by fiat. We will show below in Section 7 that both the original and

textbook Schwarzschild solutions come from the same imposed condition, g = 1, but
correspond to two different boundary conditions imposed on the solution.
While work has been done on the original Schwarzschild solution, much of the emphasis
has been on attempting to discredit the textbook solution as a solution [1421] On the other
hand, the proponents of the textbook solution view the original solution as misguided at
worst and nothing new and equivalent to the textbook solution at best [2224]; most often,
it is simply ignored. In addition, the use of gauge function terminology [25], distinguishing
between the original Schwarzschild solution and the textbook solution, or Hilberts solution,
as different gauges, provides a false sense of understanding and obscures the underlying
relationship between these two solutions.
To clarify the issues surrounding and underlying the two Schwarzschild solutions, we
begin by simply removing all assumptions and agendas in the derivation of the solution and
examine the rigorous exact solution without bias. To facilitate connections with historical
papers that have worked with the differential equation approach and to reach the widest
possible audience, we choose to work in a standard coordinate basis and with the resulting
differential equations. In Section 2 we define our coordinate basis and derive the geodesic
equations, Christoffel symbols of the second kind, the components of the Ricci tensor, and
the vacuum field equations for the spherically symmetric system without any assumptions
or imposed conditons on the coordinates. In Section 3 we provide our derivation of the most
general time-independent, spherically symmetric solution of the vacuum field equations.
3

This is the solution mentioned above that was derived independently by Combridge and
Janne. In the process of doing so, we explicitly demonstrate de Sitters discovery relating to
the number of independent field equations which underlies the Combridge-Janne solution.
While we use the Combridge-Janne solution as the basis for our analyses, we focus almost
exclusively on understanding the relationship between the original Schwarzschild solution
and Hilberts version of the Schwarzschild solution. The importance of the CombridgeJanne solution, however, goes well-beyond understanding the connections between these
two Schwarzschild solutions [26].
We show that Hilberts version of the Schwarzschild solution and the original Schwarzschild
solution are distinct solutions and can not be shown to be equivalent by means of a coordinate transformation (Section 4). In Section 5, we then discuss the issues involved in using
the g22 metric function as a radial coordinate. In Section 6 we derive a very simple but
rigorous equation relating the g22 metric function and the determinant of the metric. 1 We
then show that that equation is the groundwork for a generalization of the Schwarzschild
solution into a one-parameter family of solutions that includes both Hilberts version of
the solution and the original solution on an equal footing as special cases (Section 7). It
is explicitly shown that boundary conditions provide the basis for the individual solutions
which establishes them as separate and distinct solutions. We summarize and discuss our
results in Section 8.

Derivation of the Field Equations

We choose our coordinate basis to be a time-like coordinate and the standard spherical
spatial coordinates, (x0 = ct, r, , ). In all that follows, the symbols t, r, , and always
refer to this coordinate basis and any change of coordinate basis is explicitly stated and
utilizes typographically distinct symbols. It is well-known that the most general line element for a time-independent and spherically symmetric system in this coordinate basis
has the form [1, 2]
ds2 = A(r)(dx0 )2 B(r)dr2 C(r)d2

(2.1)

d2 = d2 + sin2 d2

(2.2)

where the three metric functions A, B, and C are functions of r only and d is the
differential solid angle. This line element is stationary (A, B, and C are independent of t),
1

The Schwarzschild condition on the determinant of the metric is not the only imposed condition that
may be used to determine the g22 metric function. A wide variety of imposed conditions may be used in
its place [26].

static (g0k = 0 for k = 1, 2, 3), and rotationally invariant (g33 = g22 sin2 and A, B, and C
are functions of |r| only). Since the limiting form of the line element for sufficiently large
distances from the source of spacetime curvature is expected to be Lorentzian
ds2 (dx0 )2 dr2 r2 d2

(2.3)

then A(r) 1, B(r) 1, and C(r) r2 in the asymptotic region. The last condition
ensures the circumference of a great circle is given by 2r and the surface area of a sphere
is given by 4r2 in the asymptotically flat spacetime.
p
We now introduce the radial function 2 R(r) = C(r) rather than choose C(r) = r2 .
The choice C(r) = r2 obscures the underlying structure of the equations and provides no
significant simplification in the solution process. In addition, it is unnecessarily restrictive.
The conventional
papproach is to caste such a choice in the language of a coordinate transformation,
p r = C(r). However, such an approach requires the new radial coordinate to
obey C(0) r < . To maintain that 0 r < means either the radial coordinate
r was not obtained from C(r), effectively resultingpin C(r) = r2 as a choice for C(r), or
p
C(0) = 0 was tacitly imposed. The condition C(0) = 0, p
however, is not a general
condition but is a specific boundary condition on the function C(r) and is appropriate
for a specific source object, the source now known as a wormhole. This has been shown to
be a specific case of a more general solution [26] which will also be demonstrated below in
Section 7.
To display obvious parallels with the standard results and field equations, we will adopt
the customary exponential forms for A(r) and B(r) [1, 2] so that we write the metric as
ds2 = e (dx0 )2 e dr2 R2 d2

(2.4)

where , , and R are functions of r only. The Lorentzian character of the metric in the
asymptotic region now means that both and vanish sufficiently far from the source
while R(r) r.
We follow the standard approach to calculating the Christoffel symbols of the second
kind and note that the equations of the geodesic lines

x
+

x x = 0

(2.5)

result from the Euler-Lagrange equations obtained from the variational problem
2

The symbols R and R(r) will always refer to the g22 metric function. To avoid confusion with the Ricci
scalar, any references to the Ricci scalar will use the contracted tensor symbol R .

F ds = 0

(2.6)

where the function F is given by


 2

+ sin2 ()
F = e (x 0 )2 e (r)
2 R2 ()
2

(2.7)

In the above equations, the overhead dot denotes differentiation with respect to the line
element variable s, x 0 = dx0 /ds, r = dr/ds, = d/ds, and = d/ds. The EulerLagrange equations resulting from this variational problem are


d F
F
=0

ds x
x

(2.8)

Calculating the partial derivatives and simplifying the resulting equations, we obtain the
geodesic equations for = 0, 1, 2, 3, respectively,

2e x
0 + r x 0 = 0
1
1
2 e RR sin2 ()
2 e RR ()
2=0
r + e (x 0 )2 + (r)
2
2

(2.9)
(2.10)

R
+ 2 r sin cos ()
2=0
R
+ 2

R
r + 2 cot = 0
R

(2.11)
(2.12)

where the prime denotes differentiation with respect to the radial coordinate r. Comparing
these equations to the geodesic equations (2.5) for = 0, 1, 2, and 3 in turn provides all of
the non-zero Christoffel symbols which we list in Table 1. In addition, the determinant of

the metric tensor, g, and the relevant term that appears in the Ricci tensor, ln g, are
also given in Table 1. These equations are the generalization of the standard results for
the geodesic equations, Christoffel symbols, and determinant of the metric by including
the g22 function R(r). It is very easy to verify that these equations revert to the standard
textbook results for R(r) = r.
The Ricci tensor may be written as

Table 1: The non-zero Christoffel symbols of the second kind obtained from the geodesic equations,

eqs. (2.9) - (2.12). We also include entries for the determinant of the metric g and ln g.

0
1





1
2

2e

1
2

e RR

e RR sin2

1
1

2
1







R
R

2
3

sin cos
3

cot

R
R

2
3

e+ R4 sin2

g = ||g ||
ln

1
2 (

+ ) + 2 ln R + 12 ln(sin2 )

R = ln

||

+
|



ln

(2.13)

For static and spherically symmetric systems, the only non-zero derivatives are those with
respect to x1 = r (contravariant coordinate index = 1) and x2 = (contravariant
coordinate index = 2). By direct calculation, the non-zero components of the Ricci
tensor are

R00

e  2 2 R 
+

+
=
2
2
2
R

R11 =

(2.14)

1  2 2 R 4R 
+

+
2
2
2
R
R

(2.15)

 + 

(2.16)

R22 = (e RR ) 1 + (e RR )

 + i
h

R33 = sin2 e RR 1 + (e RR )
= sin2 R22
2

(2.17)

All other R are identically zero and provide no additional information. Eqs. (2.14)
(2.17) provide the most general Ricci tensor components for the time-independent spherically symmetric problem.
The vacuum equations, R = 0, now yield the three general relativistic field equations

2 2 R

+
=0
2
2
R

(2.18)

2 2 R 4R

+
=0
2
2
R
R

(2.19)

+
+

(e RR ) 1 + (e RR )

 + 
2

=0

(2.20)

since the R33 = 0 equation duplicates the R22 = 0 equation. These are the timeindependent spherically symmetric vacuum field equations which explicitly contain the
8

g22 metric function R(r). Although it is often stated that, for the time-independent spherically symmetric case, R = 0 is a system of three independent equations, we wish to
re-state and emphasize that de Sitter has shown that only two of the three field equations
are independent [11]. Consequently, the field equations are not sufficient to fully determine
the solution.

Derivation of the Combridge-Janne Solution

We will now provide the solution of the vacuum field equations without any additional
assumptions or imposed conditions, such as Hilberts choice of coordinates such that

R = r or Schwarzschilds choice of coordinates such that g = 1. It will eventually


become clear that a choice of coordinates has nothing to do with such imposed conditions.
It will also become evident that the solution is quite straightforward and is not in need of
any purported simplifications.
Subtracting (2.19) from (2.18) gives
2R
=0
R

(3.1)

R 2 R
+
=0
R
R

(3.2)

+
Substituting eq. (3.1) into (2.18) yields

+ 2

Substituting (3.1) into (2.19) also produces (3.2). Substituting (3.1) into eq. (2.20) and
simplifying yields
d
(e RR2 ) = R
dr

(3.3)

So now our three independent equations are (3.1), (3.2), and (3.3). This last equation,
(3.3), may be integrated to yield
e R2 = 1

where is an integration constant. Eq. (3.1) may also be easily integrated to obtain
9

(3.4)

e+ = eC0 R2

(3.5)

where C0 is another integration constant. Now, comparing (3.5) with (3.4), we find that

e =e

C0

1
R

(3.6)

Our three equations are now (3.6) (an algebraic equation relating and R), equation (3.4)
(a differential equation relating and R), and equation (3.2) (another differential equation
relating and R).
Using eq. (3.6) to eliminate and from eq. (3.2) does not produce a differential
equation for the function R(r), it produces an identity with no conditions on the function
R(r). To see why this is so, we note that (3.2) contains only and R and their derivatives.
By direct calculation of the derivatives of (3.6), we find

R
R2





C0 R
e =e
R2
e = e C0

(3.7)
(3.8)

Taking the ratio of these equations removes the constant factor eC0 and produces


e
R /R2
 =

e
R /R2

(3.9)

Performing the differentiations in (3.9) and simplifying yields

+ 2

R 2 R
+
=0
R
R

(3.10)

which we immediately recognize as eq. (3.2). Thus, the third field equation (3.2) is
ultimately a consequence of eqs. (3.1) and (3.3); there are only two independent field
equations, not three. This provides an explicit demonstration of the result obtained by de
Sitter [11].

10

The asymptotic far-field limit determines the constant C0 to be zero and the constant
to be twice the geometric mass of the source, = 2mG = 2Gm/c2 , now known as the
Schwarzschild radius. We now have determined the constants of integration, but the metric
functions (r) and (r) remain as functionals of the g22 metric function R(r). Consequently,
having used all the available field equations, the function R(r) remains undetermined and
the solution to the vacuum field equations is not unique. It may be characterized as a
one-function family of solutions and can be summarized as

e = 1


e =

dR(r)
dr

2 

2mG
R(r)
2mG
1
R(r)

(3.11)
1

R(r) = undetermined

(3.12)
(3.13)

We do know, however, that R(r) must obey the Lorentzian asymptotic condition R(r) r
as r . As we mentioned in Section 2, this will guarantee the circumference of a
great circle in the asymptotic flat spacetime is 2r and the surface area of a sphere is
4r2 . Although R(r) has the interpretation of the radius of a circle, C = 2R(r), or of
a sphere, A = 4R2 (r), in curved spacetime, the flat space conditions must be fulfilled in
the asymptotically Lorentzian spacetime only.
To cast these results in the notation utilized by de Sitter, Combridge, and Janne, we
define the function (r) by
R = re/2

(3.14)

ds2 = e c2 dt2 e dr2 e r2 d2

(3.15)

The metric may now be written as

This is the form of the metric originally established by de Sitter [11] and used by Combridge [12] and Janne [13]. The undetermined nature of R(r) means that the function (r)
is undetermined.
Previously, the standard approach to removing the indeterminancy of the solution involving , , and has been to assume some relation among the functions , , and
[11, 18]. For instance, choosing = 0 is equivalent to the symmetry condition that
R(r) = re/2 which produces the isotropic metric
11

ds2 = e c2 dt2 e dr2 + r2 d2

mG  4
R2 
=
1
+
r2
2r
2
1 2mr G

e =
2
1 + 2mr G

e = e =

(3.16)
(3.17)

(3.18)

We note that this metric has not been obtained by means of a coordinate transformation to
isotropic coordinates [1, 27] or any other set of coordinates, the coordinate basis remains
the original spherical spatial coordinate basis introduced in Section 2. It has been obtained
by simply exploiting the extra degree of freedom provided by the indeterminate nature of
R(r) to produce the metric with the desired symmetry. Such an ab initio approach for
the isotropic metric was advocated nearly 100 years ago by Eddington [28]. We should
point out, however, that the condition R(r) = re/2 is valid only for e/2 0: it must be
supplemented by another condition if e/2 < 0 is allowed [27].
Another example is provided by imposing the condition + 2 + = 0. Although de
Sitter identified this condition as corresponding to Einsteins condition of choice [11], it
more accurately produces the condition R2 R4 = r4 which may be used as the basis for deriving the original Schwarzschild solution. This provides an indication that Schwarzschilds
original solution and the textbook solution do arise from the same condition. As a final
example, Hilberts version of the Schwarzschild solution also immediately results from simply choosing = 0 so that R(r) = r. While other choices are admissable, it must be
stressed that any such condition is arbitrary and amounts to nothing more than choosing
a metric function by decree; such a procedure provides no fundamental determination of
the function R(r).

The Two Schwarzschild Solutions

The Combridge-Janne solution reduces to the original Schwarzshild solution and Hilberts
version of the Schwarzschild solution for Rs (r) = (r3 + (2mG )3 )1/3 and RH (r) = r, respectively. We will now demonstrate that these two versions of the Schwarzschild solution are
physically distinct and are not reducible to each other by means of a coordinate transformation. First, we note that it is not possible to write the two metrics in terms of different
coordinate bases. For example, if we write

ds =




2mG 1 2
2mG 2 2
c dt 1
dr r2 d2
1
r
r
12

(4.1)




2mG 2 2
2mG 1 2
2
ds = 1
c dt 1
dr Rs2 d
Rs
Rs
2

(4.2)
Rs (r) = r3 + (2mG )


3 1/3

where Rs = Rs (r), the two spherical coordinate bases (r, , ) and (r, , ) are related to
two corresponding rectangular coordinate bases (x, y, z) and (x, y, z). Because the range
of the two sets of spherical coordinates are identical, the corresponding ranges of the rectangular coordinate bases are also identical. At most, the two coordinate systems can be
distinguished by a translation and a rotation, but there can be no deforming transformations that may alter the range of the coordinates. Consequently, there is no way to
distinguish between the coordinate bases (r, , ) and (r, , ) so that r = r without loss
of generality. In fact, we show explicitly in Section 7 that the two Schwarzshild solutions
are obtained in the same coordinate system, (t, r, , ), the basis that was introduced in
Section 3.
Second, we note that by using the chain rule, we may convert the line element (4.2)
from the Combridge-Janne form to the conventional textbook form

ds2 =




2mG 1 2
2mG 2 2
c dt 1
dRs Rs2 d2
Rs
Rs

(4.3)

While the chain rule has converted the g22 function Rs (r) into the radial coordinate, the
only function that can be used in conjunction with the chain rule to produce the line
element (4.3) for the original Schwarzschild solution is the function defined in (4.2). The
function in (4.2) no longer provides the g22 function as a function of the radial coordinate
r, but it does provide the necessary relationship between the two radial coordinates r and
Rs in metrics (4.1) and (4.3), respectively:

1/3
Rs = r3 + (2mG )3

(4.4)

We point out that eq. (4.4) does not function as the coordinate transformation that puts
one metric in the form of the other: using (4.4) in the textbook metric (4.1) does not
produce the original metric (4.3) and using (4.4) in the original metric (4.3) does not
produce the textbook metric (4.1).
It is clear that (4.3) has the form of the textbook Schwarzschild metric. It is also clear
that Rs 2mG since eq. (4.4) shows that Rs (0) = 2mG . Thus, it is easy to misinterpret
13

(4.3) as corresponding to the textbook solution outside of the event horizon. To clarify
the issue, we note that the textbook solution has an event horizon at r = 2mG which
corresponds to Rs = (2)1/3 (2mG ). The line element (4.3) has no event horizon at that
radius, it is perfectly well-behaved there. In fact, it is perfectly well-behaved everywhere
except for a singularity at Rs = 2mG . But Rs = 2mG corresponds to r = 0 not to r = 2mG .
So, delineating the regions relative to the site of the textbook event horizon, the region
exterior to the textbook event horizon corresponds to the ranges

r > 2mG
(4.5)
Rs > (2)

1/3

(2mG )

and the region interior to the textbook event horizon corresponds to the ranges

0 r < 2mG

(4.6)

2mG Rs < (2)1/3 (2mG )


Consequently, the metric (4.3) covers the entire spacetime, both inside and outside of the
textbook event horizon with no complications or subtleties of any kind. The textbook
metric (4.1) has an event horizon at r = 2mG and a temporal singularity at r = 0 since
the radial coordinate r is timelike inside the event horizon. The original metric (4.3) has
no event horizon corresponding to the textbook event horizon and has a spatial singularity
and coinciding event horizon at Rs = 2mG or r = 0. These are two completely different
metrics describing two completely different physical configurations.
We note in passing that the singularity of the metric (4.3) at Rs = 2mG appears to be
at a spherical surface. This, however, is an artifact of using the g22 function Rs (r) as the
radial coordinate. This is a problem of interpretation that does not appear when using the
original coordinate basis (t, r, , ). The minimum value of Rs (r) simply means that the
g22 metric function has a non-zero minimum at r = 0, it does not imply a minimum value
for the radial coordinate r different from r = 0. The singularity at Rs = 2mG does not
indicates a surface singularity, it corresponds to a point singularity at r = 0. Other issues
of interpretation and meaning related to using the general R(r) as the radial coordinate
are discussed below in Section 5.

14

Finally, we offer one last and perhaps more straightforward approach to demonstrating
the inequivalence of the textbook metric and the original metric. Given Hilberts version
of Schwarzschilds solution [1, 2],

ds =




2mG 2 2
2mG 1 2
1
c dt 1
dr r2 d2
r
r

(4.7)

where 0 r < , and the original Schwarzschild solution


2

ds =




2mG 2 2
2mG 1 2
1
c dt 1
dRs Rs2 d2
Rs
Rs

(4.8)

where Rs (0) = 2mG Rs < , we ask What is the coordinate transformation that
converts the textbook solution and the original Schwarzschild solution into each other? If
we denote this transformation by Rs = f (r), then the original Schwarzschild line element
(4.8) becomes

ds =




2mG 1 2 2
2mG 2 2
c dt 1
f dr f 2 d2
1
f
f

(4.9)

If we now demand that this be identical to Hilberts line element, this requires f to satisfy
three equations,
2mG
2mG
=1
f
r




2mG 2
2mG
1
f = 1
f
r
1

f 2 = r2

(4.10)
(4.11)
(4.12)

The only solution for f (r) is f (r) = r which requires Rs = r. Since the original Schwarzschild
solution has Rs3 = r3 + (2mG )3 , this can only be satisfied by mG = 0. The original
Schwarzschild solution and Hilberts solution are identical only for the case in which there
is no source and spacetime is strictly Lorentzian everywhere.

15

R(r) as a Coordinate Transformation

The discussion in Section 4 suggests one way of removing the indeterminacy associated
with the g22 function R(r) is to treat R(r) as a coordinate. In other words, define a
coordinate transformation from the original spherical radial coordinate r to a new radial
coordinate r by means of the function r = R(r). While it is possible and, at times, very
convenient to do so, we hasten to point out that a great deal of information may be lost by
treating it in this manner and a number of problems of interpretation may be introduced
as well. The form of the Combridge-Janne solution shows that the function R(r) contains
information about the source object. Specifically, in the limit as r 0, R(r) describes the
approach to the source. However, treating R(r) as a coordinate then requires R(0) to be
treated as an adjustable parameter which will also pose problems of interpretation for the
parameter R(0). In addition, the boundary value R(0) also is specific to the source object
and does not take on a single value for all sources: R(0) parametrizes the function R(r)
and each value of R(0) provides a different function R(r).
Although r = R(r) may be given an interpretation as a radial coordinate, it can not
be identified with the spherical radial coordinate r, but is clearly a mapping of the radial
coordinate r. In particular, the radial coordiate r lies in the range 0 r < while the
new coordinate r = R(r) has the range

R(0) r <

(5.1)

where, as in Section 4, we have used the asymptotic behavior, R(r) r as r , to


identify the infinite upper bound on r = R(r). If the function R(r) has not been determined,
the parameter R(0) can not be obtained from a boundary condition on R(r) and must then
be treated in a phenomenological manner. This phenomenological treatment must be based
upon a relationship between the lower bound R(0) and the nature of the source object. This
relationship will not have the transparent characteristics of a boundary condition on the
function R(r) and must have the character of a model-dependent statement. Consequently,
treating r = R(r) as an independent radial coordinate replaces an undetermined function
by at least one undetermined adjustable parameter. If there are any other values of r
that identify characteristic features of spacetime, then those values will also have to be
treated as undetermined adjustable parameters. For example, the textbook event horizon
at r = 2mG corresponds to R(2mG ), but R(2mG ) is unknown unless R(r) is known and so
it must also be treated in a phenomenological manner.
We also stress that the radial coordinate r is a location marker, it identifies the radial
location of a point in space-time. The function R(r), in its role as the g22 metric function,
provides the distance in curved space-time along an arc of a great circle, R(r), where
is the angle defined by the arc. Unless space-time is flat, there is no reason to expect the
16

radial location marker r to provide the distance in curved space-time along an arc of a great
circle. This means, as noted above, even if using R(r) as the radial coordinate marker,
R(r) can not be identified with the standard radial marker r except in the asymptotically
flat Lorentzian space-time. As a metric function, R(r) reflects the twists and turns of
curved space-time while r, as a radial location marker, does not. Thus, using R(r) as an
independent coordinate means R(r) has two separate and distinct roles, one as the radial
location marker and one as a measure of distance in curved space-time, and any information
that R(r) may have about the properties of curved spacetime will be lost if it is treated
exclusively as a coordinate.
Nevertheless, if we insist on treating R(r) as a coordinate transformation, we can examine the geodesic equations and their implementation in determining physical characteristics
such as planetary orbits, the trajectory of a light ray in space-time, and so on in terms
of R(r). The geodesic equations for the time-coordinate (eq. (2.9) and the two angular
coordinates (eqs. (2.11) and (2.12)) are
d 0
(e x ) = 0
ds

(5.2)

d
= R2 sin cos ()
2
(R2 )
ds

(5.3)

d
=0
(R2 sin2 )
ds

(5.4)

In place of the radial geodesic equation, we use the line element itself to obtain
1 = e (x 0 )2 e R2 r 2 R2 (2 + sin2 2 )

(5.5)

This last equation, for the case of a light trajectory, is replaced by


0 = e (x 0 )2 e R2 r 2 R2 (2 + sin2 2 )

(5.6)

The time-coordinate and angular coordinate equations are the standard ones. We fix the
trajectory in the x-y plane by the constant solution for , = /2 which then yields the
standard results
e x 0 = L0 = constant

17

(5.7)

R2 = L = constant

(5.8)

R2 sin2 = 0

(5.9)

The remaining equation, however, is altered due to the presence of R . We note that R

= R r we may write
only occurs in conjunction with r,
so using the chain rule r = R(r)
equation (5.5) and (5.6) as
2

1 = e (x 0 )2 e r r2 (2 + sin2 2 )

(5.10)

2
0 = e (x 0 )2 e r r2 (2 + sin2 2 )

(5.11)

where (5.11) is appropriate for a light trajectory. Now the standard analyses [1] of trajectories and Keplerian orbits follow with the exception that the radial coordinate r = R(r)
has replaced the radial coordinate r as the measure of radial distance. In other words,
rather than introducing the variable u = 1/r, the equations invite the introduction of the
variable u = 1/r. In particular, measures of distance from the origin r = 0 translate to
measures from r = R(0). This has no effect on the calculation of the perihelion shift of
Mercury and, in the case of the deflection of a light ray, it leads to the interpretation of
R(0) as the distance of closest approach to the origin. This suggests the possibility of a
more general interpretation that relates R(0) to the geometrical size of the source object.

An Equation for R(r)

It was explicitly shown in Section 3 the the field equations do not determine the metric
function g22 = R2 (r). Consequently, we would like to derive an equation for R(r) that
lies within the theoretical framework of general relativity [26]. The standard textbook
presentation of Hilberts version of the solution [1, 2], R(r) = r, specifies the metric function by decree and can provide no guidance for deriving such an equation. The postulated
relations among the metric functions , , and (see the end of Section 3) [11, 18] have
no underlying theoretical basis so they also provide no guidance for deriving an equation
for R(r). But Schwarzschilds original solution [68] imposed the condition that the determinant of the metric have a specified value. More exactly, he introduced coordinates

that permitted the imposition of the condition g = 1 for all values of the coordinates.
Consequently, we seek a functional relationship between the determinant of the metric

18

g ||g || = e+ R4 sin2

(6.1)

and the g22 metric function R(r). Using eq. (3.5) in (6.1), recalling that C0 = 0, and
taking the square root yields

g = R R2 sin

(6.2)

Simple rearrangement then provides an equation for R(r),

g
1 dR3
=
R R =
3 dr
sin
2

(6.3)

Eq. (6.3) is a fundamental relationship between the g22 metric function R(r) and the
determinant of the metric expressed in the coordinate basis introduced in Section 2. Since

the left hand side of eq. (6.3) is a function of r only, g, when expressed in spherical
coordinates, must be linear in sin . This will guarantee that the right hand side of eq.
(6.3) yields a function of r only.
Eq. (6.3) shows that if R(r) is specified, then the determinant of the metric tensor will
be uniquely determined. Conversely, if the determinant of the metric tensor is not specified
by some condition, then (6.3) shows that the metric function R(r) remains undetermined.
Although eq. (6.3) is a rigorous consequence of the spherically symmetric vacuum field
equations, we emphasize that an arbitrary specification of the determinant of the metric
still produces an arbitrarily specified metric function R(r).
We also emphasize that there is no obligation to make use of (6.3) to determine
R(r) [26]. Using an alternative condition to determine R(r) does not conflict with (6.3),
it simply provides a means of specifying the determinant of the metric. Any imposed condition for determining R(r) may be used as long as the resulting solution for R(r) obeys
the asymptotic condition R(r) r far from the source. As a result, the imposed auxiliary

condition and the corresponding solution for R(r) provide a determination of g from
eq. (6.3).

The Generalized Schwarzschild Solution

Since our current focus is on understanding the relationship between the original
Schwarzschild solution and Hilberts version of the solution, we use eq. (6.3) as a means
to generalize Schwarzschilds original solution. First, we note that the determinant of the
19

metric tensor is not a scalar but a scalar density of weight W = 2 and, consequently, the
square root of the determinant of the metric is a scalar density of weight W = 1,



X 1 p

g =
g
X

(7.1)

where |X/X| is the Jacobean for the coordinate transformation that connects the barred
coordinate system (with determinant denoted by g) to the unbarred coordinate system
(with determinant denoted by g). If, in a specific coordinate system, the determinant of
the metric tensor has a particularly simple form, then equation (7.1), in conjunction with
eq. (6.3), can then be used to determine the function R(r).
As an illustration, we now make an explicit connection with Schwarzschilds original
solution by using Schwarzschilds coordinate transformation [68]

x0 = x0 = ct
r3
(x1 )3
=
3
3
x2 = cos(x2 ) = cos
x1 =

(7.2)

x3 = x3 =

The Jacobean for this transformation is




X
1 2
2 1
2
1


X = ((x ) sin x ) = (r sin )

(7.3)

The barred coordinate system is the system used to impose the condition on the determinant of the metric tensor, g = 1. Consequently, the transformation for the square root
of the determinant of the metric gives

g = ((x1 )2 sin x2 )

p
p
g = (r2 sin ) g = r2 sin

(7.4)

Eq. (7.4) shows that working in coordinates which permit the condition g = 1 is

equivalent to working in the original spherical coordinate basis and requiring g =


r2 sin . In the spherical coordinate system, the determinant g is related to R(r) by (6.3)
so that using eq. (7.4) yields
20

dR3
= 3r2
dr

(7.5)

Eq. (7.5) which determines R(r) was not produced by means of a choice of coordinates. It
is due to the additional auxiliary condition, eq. (7.4), that has been imposed to determine
the g22 function. The choice of coordinates only affects the form of the condition, whether

it takes the form of g = 1 in the coordinates introduced by Schwarzschild or g =


r2 sin in the standard spherical coordinate basis introduced in Section 2.
As shown in Section 4, the spherical coordinate basis is indistinguishable from the
coordinate basis used in Hilberts derivation of the Schwarzschild solution, the difference
in the solutions is in the respective g22 functions not in the coordinates. As discussed in
Section 2, since Hilberts derivation of the solution does not specify a source object, it does
not impose a condition on coordinates, it simply imposes the condition R(r) = r. Likewise,

in Schwarzschilds original solution, the condition g = 1 was not automatically satisfied


by means of the coordinate transformation given by (7.2), the condition was separately
and explicitly imposed. In fact, in the coordinates he used, Schwarzschild could have

imposed some other value on g, the coordinates did not restrict the value in any way.
While we have chosen to impose condition eq. (7.4) to connect with the original and
textbook versions of the Schwarzschild solution, we stress that there is no compelling
reason to choose the imposed condition eq. (7.4). As stated at the end of Section 6, any
other imposed condition is equally viable as long as the solution of the imposed condition
gives the asymptotic behavior R(r) r as r approaches the asymptotically flat Lorentz
spacetime [26].

The imposed condition, g = 1 or g = r2 sin has resulted in the differential


equation (7.5). This equation can be easily integrated to give
Rr0 = (r3 + r03 )1/3

(7.6)

where r0 is an integration constant 3 which parametrizes the solution. Using (7.6) for
Rr0 (r) together with the solutions for and constitutes the generalized Schwarzschild
solution [26] and provides the basis for both the original Schwarzschild solution (r0 = 2mG )
and Hilberts version of the Schwarzschild solution (r0 = 0). 4
The functions given by (7.6) for the two cases r0 = 2mG and r0 = 0 are shown in
Figure 1. Clearly the differences between the two functions are restricted to the small r
region, r . 4mG . Although the Schwarzschild Rr0 (r) is only somewhat more complicated
than the Hilbert version, the impact of Schwarzschilds Rr0 (r) on the overall solution is
3

Schwarzschilds original notation used the parameterp = r03 [68].


Note that setting r0 = 0 is equivalent to Rr0 (0) = C(0) = 0 which provides a direct connection to
Hilberts original approach.
4

21

Schwarzschild
Hilbert, space-like r
Hilbert, time-like r

Rr0 (r)
2mG

0
0

r
2mG

Figure 1: The original Schwarzschild Rr0 (r) with r0 = 2mG compared to Hilberts function Rr0 (r)
with r0 = 0. The change of r to the time-like coordinate for the Hilbert solution is reflected in the
change of color of the plot at r = 2mG .

dramatic. The overriding simplification of the original Schwarzschild solution is the lack of
a role-swapping transition between the time-like coordinate and the radial coordinate, the
radial coordinate and the function Rr0 (r) for the original Schwarzschild solution maintain
their space-like character for all r. This is simply a reflection of the differences in complexity
of the corresponding source objects.
Although the differences between the Schwarzschild and Hilbert Rr0 (r) functions are
significant for small r, they are not astonishing. However, the resulting differences in the
other metric functions are more visually notable. Figure 2 shows the differences between
the Schwarzshchild g00 function and the Hilbert g00 function. For r 2mG where the
Hilbert g00 function changes sign, the Schwarzschild function remains postive semi-definite
and forms a shallow well. While also restricted to the small r region, the differences between
the two functions are considerable larger than the corresponding differences between the
g22 functions. The distinction between the finite potential spatial well and the temporal
singularity at r = 0 is particularly striking.
Figure 3 shows the g11 functions for the original Schwarzschild and Hilbert solutions.
The dissimilarity between the coordinate singularity at r = 2mG and the physical singularity at r = 0 displays the dramatic simplification of interpretation for the original
Schwarzschild solution. As the parameter r0 is permitted to increase from zero to 2mG ,
the vertical asymptote at r = 2mG slides to the left until it coincides with the vertical
asymptote at r = 0 for r0 = 2mG . The distinction between the purely space-like role of
22

2
Schwarzschild
Hilbert, space-like r
Hilbert, time-like r

4
0

r
2mG

Figure 2: e for the original Schwarzschild and Hilbert solutions. The change of r to the time-like
variable for r < 2mG for the Hilbert solution is indicated by the change of color in the plot.

20

10
0
10

Schwarzschild
Hilbert, space-like r
Hilbert, time-like r

20
0

r
2mG

Figure 3: e for the original Schwarzschild and Hilbert solutions. The change of r to the time-like
variable for r < 2mG for the Hilbert solution is indicated by the change of color in the plot.

23

the original Schwarzschild and the composite space-like and time-like roles of the Hilbert
solution highlight the overall simplicity of the point mass solution compared to that of a
wormhole.
As we mentioned at the end of Section 3, the metric function condition + 2 + = 0
produces the square of eq. (7.5). In this language, + 2 + is directly related to the
determinant of the metric and not its square root,
g = e+2+ r4 sin2

(7.7)

Nevertheless, this clearly shows that postulating + 2 + = 0 is equivalent to requiring


g = r4 sin2 which leads to eq. (7.4), eq. (7.5), and eq. (7.6) for the appropriate choice
of sign for the square root of (7.7). It also provides the connections among Einsteins
condition of choice, Schwarzschilds original solution, and the determinant of the metric.
From the point of view of Hilberts approach to the Schwarzschild solution, Rr0 (r) is
treated as a coordinate and not as a metric function. Eq. (7.5) provides the equation relating the original radial coordinate r to the new radial coordinate Rr0 and eq. (7.6) provides
the explicit functional dependence of Rr0 (r) on the original radial coordinate r. The solution for Rr0 (r), eq. (7.6), was obtained from a differential equation and the conventional
method for determining constants of integration is to supply boundary conditions on the
solution. To obtain Hilberts
psolution, the constant r0 is required to be set equal to zero,
which means that Rr0 (0) = C(0) = 0. While there is nothing ill-defined by such a choice,
it must be admitted that such a choice corresponds to no particular well-defined boundary
condition on Rr0 (r). In fact, the condition is imposed before the field equations are even
established. Typically the boundary condition reflects the nature of the source object by
means of the behavior of the function at a boundary or as the source object is approached.
But in the derivation of Hilberts solution, the nature of the source is never mentioned and
the solution proceeds as if the nature of the source object is irrelevant or that the solution
is universal and appropriate for all sources. The only condition that may be stated as
the boundary condition for Hilberts solution is the condition that Rr0 (r) = r everywhere
which makes no obvious statement about the nature of the source object.
By contrast, in his original 1916 paper, Schwarzschild was explicitly concerned with a
unique solution to the gravitational field equations, free of any undetermined parameters,
that was appropriate for a point mass. Consequently, he determined the parameter r0
by requiring the metric functions to be continuous everywhere except at the origin, the
location of the point mass [68]. This requires the parameter r0 to be the Schwarzschild
radius, r0 = 2mG . It is clear, however, that any value of r0 greater than the Schwarzschild
radius will satisfy the requirement that the metric functions be continuous everywhere,
including the origin r = 0. This shows that r0 2mG provides physically meaningful
solutions but the solutions that are regular at r = 0 are solutions for objects of a finite
size, not for a point mass. This suggests that the parameter r0 may be allowed to take on
24

any value in the range r0 2mG depending on the particular source object of the problem.
In addition, by appealing to the Hilbert solutions discontinuity in the g11 function at
r = 2mG as motivation, if we allow the parameter r0 to also take on values in the range
r0 < 2mG , we obtain a far-reaching generalization of Schwarzschilds solution by relaxing
the continuity condition on the metric functions and permitting any real value for the
parameter r0 . This simply means that the Schwarzschild parameter r0 is left as a free
parameter to be determined by a boundary condition and eq. (7.6) characterizes the
generalized Schwarzschild solution as a one-parameter family of solutions [26].
This generalized solution contains both the original Schwarzschild solution (r0 = 2mG )
and Hilberts version of the solution (r0 = 0) as special cases. It also illustrates the
relationship between the two versions of the Schwarzchild solution: they are siblings in
the one-parameter family of solutions and are not equivalent to each other and related by
means of a coordinate transformation. From elementary real analysis, it is known that two
members of a one-parameter family of functions are equivalent if and only if the familyparameters for the two members are equal. In the context of the current discussion, this
requires the value of r0 for the original solution and the value of r0 for Hilberts solution
to be equal. In other words, mG = 0, as we found in Section 4 by another method.
The function Rr0 (r), eq. (7.6), has been determined by a first order differential equation, eq. (7.5), which is not linear in Rr0 (r) but is linear in Rr30 . Any constants in that
solution are determined by boundary conditions for the given physical configuration. The
boundary condition at infinity, the asymptotic condition that Rr0 r in the asymptotic
Lorentzian spacetime, is satisfied by (7.6) for any finite value of r0 . The appropriate boundary condition for small r depends on the nature of the source object. For Schwarzschilds
point mass, the metric should be regular everywhere except at the site of the point mass,
r = 0. This requires r0 = 2mG as Schwarzschild found. If, however, the object is not
a point mass, for example if it is some object of finite extent, then the metric should be
regular everywhere that it applies. For example, by choosing r = 0 to correspond to the
surface of the object, then the parameter r0 may be determined by specifying the potential
e = 1 2mG /Rr0 (0) = 1 2mG /r0 on the surface and the solution is indeed regular
everywhere including at the surface of the source object. This also has the advantage of
completely avoiding the difficult issues that arise with the use of piece-wise continuous
functions in the solutions of nonlinear differential equations. In addition, if the source is
some type of exotic object, such as a wormhole, then r0 < 2mG will produce the necessary
3 = 8m3 r 3 , with Hilberts
coordinate singularity and corresponding event horizon at reh
0
G
5
source object described by r0 = 0. The relation of the boundary condition at small r
to the source of gravitational spacetime curvature suggests that the boundary condition
Rr0 (r) = r for all r was not originally connected to the nature of the source in any way.
From the historical perspective, it is clear that the condition Rr0 (r) = r was applied for
5
The function Rr0 (r) = (r3 |r0 |3 )1/3 is negative for r < |r0 | which may pose difficulties of interpretation.
Consequently, this may require r0 0.

25

extraneous reasons, a purported simplification, and was not motivated by the identification
of the source object as a wormhole. It was the full explication of Hilberts version of the
solution by means of alternate coordinate systems that provided a self-consistent picture
of the exotic wormhole as the physical source for Hilberts solution [1, 2, 5].
The continued use of the coordinate basis (t, r, , ) inside the textbook event horizon at
r = 2mG shows that the radial and time coordinates swap roles, the radial coordinate r is
time-like and the time coordinate t is space-like. This led to the view that the coordinates
for the textbook solution, termed the Schwarzschild coordinates, were pathological [2].
The transformation to isotropic coordinates expressed the textbook metric in a form that
had no event horizon and showed no such role swapping [1]. However, it has been shown
that the original transformation to isotropic coordinates is valid only outside the textbook
event horizon at r = 2mG and must be supplemented by another time-dependent transformation inside the event horizon [27]. The time-dependence of the textbook wormhole
solution emerges whether examined in Kruskal-Szekeres coordinates [2], isotropic coordinates [27], or simply continued use of the original coordinate basis (t, r, , ). This, in
fact, is the fundamental significance of the invariant constructed from the acceleration
associated with the time-like Killing vector [19]: an event horizon cannot be eliminated
by means of a coordinate transformation. An event horizon is the site at which a metric
exchanges its time-invariance for spatial invariance. This underlies the basic physical distinction between the two Schwarzschild metrics: the textbook Schwarzschild solution, the
wormhole solution, is time-independent outside of the event horizon but is time-dependent
inside the event horizon; the original Schwarzschild solution, the point mass solution, is
time-independent everywhere since the event horizon at r = 0 has no interior.
It should be clear by now that the original Schwarzschild solution and Hilberts version
of the Schwarzschild solution do not conflict with each other, neither solution excludes
the other, but they result from two different values of the parameter r0 in the one-family
parameter of solutions described by eq. (7.6). They describe the fields of two distinctly
different source objects. The solution for a point mass, as Schwarzschild provided, is one
problem; the solution for a wormhole, as Hilbert provided, is a completely separate and
distinct problem. These two solutions are not mutually exclusive nor are they reducible to
each other. In Section 4 we discussed the distinction of the two versions of the Schwarzschild
solution by means of an attempt to relate the two solutions by a coordinate transformation.
That attempt resulted in an explicit instance of the result known from elementary real
analysis: two members of a one-parameter family of functions are equivalent if and only if
the values of their respective parameters are equal. Here we have seen that the physical
distinction between these two solutions is reflected by their two different values of r0 . These
distinct values for r0 result from two different boundary conditions that have been imposed
on the solution for Rr0 (r), eq. (7.6), and these two different boundary conditions, in turn,
are conveniently interpreted as due to the two distinct sources of the gravitational field.
It should also be clear by now that Hilberts version of the solution need not be viewed as
the result of an error, as initially maintained by Abrams [1416], but is simply the result of
26

a certain value for the parameter r0 , r0 = 0, corresponding to the boundary condition that
Rr0 (r) = r for all r. This is more or less a geometrical condition that appears to make no
explicit comment on the nature of the source of the field. It was left to later investigations
to determine the nature of the source and to identify it as a wormhole in spacetime. On
the other hand, the original Schwarzschild solution was specifically designed to provide a
solution that was regular everywhere in spacetime except at r = 0, which clearly points to a
point mass source object. Hilberts oblique comment 6 that Schwarzschilds translation of
coordinates was ill-advised suggests that he missed two essential points of Schwarzschilds
original solution: (1) Schwarzschild did not translate coordinates, his functional form for
Rr0 (r) was the result of an imposed condition on the determinant of the metric, and (2)
while Schwarzschild could have chosen r0 = 0, the value r0 = 2mG was required to describe
the field for a point mass source. We emphasize again that these two different values of the
parameter, r0 = 2mG and r0 = 0, reflect two different boundary conditions that have been
imposed on the solution for R(r), one for a point-mass object and one for an object with
no classical analogue. Both solutions are equally viable and do not exclude each other but
describe the fields due to two entirely different and inequivalent source objects.

Conclusion

We have provided a pedagogically sound derivation of the Combridge-Janne solution, a


one-function family of solutions that is conveniently characterized by the g22 metric function. Following Schwarzschilds original approach, we then imposed the auxiliary condition

on the determinant of the metric in the form g = r2 sin to produce a simple but rig
orous differential equation that relates the the metric function R(r) to g. The solution
of this differential equation is the generalized Schwarzschild solution,

ds =




2mG
2mG 1 2
2 2
1
c dt 1
Rr0 (r)dr2 Rr20 (r)d2
Rr0 (r)
Rr0 (r)
(8.1)

3 1/3

Rr0 (r) = r3 + r0


where r0 is an integration constant that is determined by a boundary condition on the


function Rr0 (r).
The solution (8.1) produces Schwarzschilds original solution by imposing the requirement that the metric functions are continuous everywhere except at r = 0. This solution
6

See the footnote at the end of Appendix B in Ref. [4]

27

is paired with the point mass source. Eq. (8.1) also produces the textbook solution by
imposing the geometrical condition that Rr0 (r) = r everywhere. This condition has no obvious connection to a particular source object, but, as is well-known, the solution is paired
with the wormhole solution. We have demonstrated, on both mathematical and physical
grounds, that these two solutions are distinct and inequivalent and provide two distinct
and inequivalent metrics.
We note that (8.1) contains not just the textbook wormhole solution, but an entire
family of wormhole solutions. The solutions for which 0 r0 < 2mG describe wormhole
solutions of varying size, with the textbook solution corresponding to the largest wormhole.
We also note that the generalized Schwarzschild solution accomodates finite-size objects,
objects for which the potential e = 12mG /Rr0 (r) = 12Gm/(cRr0 (r)) may be specified
on a surface by choosing r = 0 to be that surface. The point mass solution can then be
seen as the intersection of the solution for objects of finite size (in the limit of zero size)
and the wormhole solutions (in the limit of zero wormhole size).
The solution (8.1) clearly shows that there is no pathological character [2] to be
associated with the coordinate basis (t, r, , ). It is the particular g22 function that is
responsible for the coordinate singularity and event horizon of the textbook solution. Of
course, this responsibility ultimately belongs to the boundary condition that produces the
completed form of Rr0 (r) and reflects the nature of the source object. If the solution is
considered to have a pathological character, that is because the source requires that
pathological nature.
The sibling relationship between different members of the generalized Schwarzschild
solution reflects the difference between the different source objects that correspond to those
solutions. Just as no gauge relationship converts one charge distribution into another, there
is no gauge relationship that converts one gravitational source object, such as a point mass,
into another source object, such as a wormhole. There is no necessary relationship among
the different imposed conditions that may be used to determine Rr0 (r) and there is no
necessary relationship among the different boundary conditions that may be imposed to
determine the constant of integration for a specific imposed condition [26].
The invariant norm a a associated with the time-like Killing vector [19] has a divergence for Hilberts solution at r = 2mG . This value of the radial coordinate is precisely the
radius for which the g11 function is singular and the time-like coordinate and the radial
coordinate exchange roles. This exchange of coordinate roles reflects the exchange of symmetries that occurs at that particular value of the radial coordinate. The event horizon
at that radius acts like a time-membrane or filter. Traversing the event horizon from
points outside the event horizon accesses past times with no possible reconnection to future
times. These problematical aspects of the wormhole solution are not due to pathological
coordinates but are distinguishing features of Hilberts g22 metric funcition, Rr0 (r) = r
and are characteristic of wormhole solutions. These features are completely foreign to
Schwarzschilds original solution since he designed his solution to deliberately avoid such
issues; such features are not characteristic of the point mass solution. In the light of this
28

discussion, it seems more accurate and more appropriate to reserve Schwarzschilds name
for the original solution and to designate the textbook solution as the Droste-Weyl-Hilbert
solution or more simply as the Hilbert solution [4].
References
[1] R. Adler, M. Bazin, and M. Schiffer, Introduction to General Relativity, Second Edition
(McGraw-Hill, New York, 1975).
[2] C.W. Misner, K.S. Thorne, and J.A. Wheeler, Gravitation (Freeman, San Francisco,
1973).
[3] D. Hilbert, Nachr. Ges. Wiss. Gottingen, Math. Phys. Kl., 53 (1917). Submitted 23
Dec. 1916.
[4] S. Antoci, David Hilbert and the Origin of the Schwarzschild Solution in Meteorological and Geophysical Fluid Dynamics, edited by W. Schroder, (Science Edition,
Bremen, 2004). Also arXiv:physics/0310104 v1.
[5] M.D. Kruskal, Maximal Extension of Schwarzschild Metric, Phys. Rev., 119, 1743
(1960).

[6] K. Schwarzschild, Uber


das Gravitationsfeld eines Massenpunktes nach der Einsteinschen Theorie, Sitzungsber. Preuss. Akad. Wiss., Phys. Math. Kl., 189, (1916).
[7] K. Schwarzschild, On the Gravitational Field of a Mass Point According to Einsteins
Theory, trans. by S. Antoci and A. Loinger, Gen. Relativ. Gravit. 35, 951 (2003).
Also arXiv:physics/9905030v1 (1999).
[8] K. Schwarzschild On the Gravitational Field of a Point-Mass, According to Einsteins
Theory, trans. by L. Borissova and D. Rabounski, Abra. Zel. Jour. 1 (2008).
[9] C. L. Pekeris, Gravitational field of a charged mass point, Proc. Natl. Acad. Sci.
USA, 79, 6404 (1982).
[10] A. Einstein, Annalen der Physik, 49 (1916). English translation in H.A. Lorentz, A.
Einstein, H. Minkowski, and H. Weyl, The Principle of Relativity, trans. by W. Perrett
and G. B. Jeffery, (Dover, New York, 1952).
[11] W. de Sitter, On Einsteins Theory of Gravitation, and Its Astronomical Consequences (First Paper), Month. Not. R. Astr. Soc., 76, 699 (1916).
[12] J.T. Combridge, Phil. Mag., 45, 726 (1923).
[13] H. Janne, Bull. Acad. R. Belg., 9, 484 (1923).
29

[14] L. S. Abrams, Alternative Space Time for the Point Mass, Phys. Rev. D, 20, 2474
(1979).
[15] L.S. Abrams, Black Holes: The Legacy Of Hilberts Error, Can. J. Phys., 67, 919
(1989). Also arXiv:gr-qc/0102055 v1.
[16] L.S. Abrams, The Total Space-Time of a Point-Mass When 6= 0, and Its Consequences for the Lake-Roeder Black Hole, Physica A, 227, 131 (1996). Also arXiv:
gr-qc/0102053.
[17] L.S. Abrams, The Total Space-Time of a Point Charge and Its Consequences for
Black Holes, Int. J. Theor. Phys., 35, 2661 (1996). Also arXiv:gr-qc/0102054 v1.
[18] S. Antoci and D-E Liebscher, Reconsidering Schwarzschilds Original Solution, Astron. Nachr., 322, 137 (2001). Also arXiv:gr-qc/0102084 v2.
[19] S. Antoci and D-E Liebscher, Reinstating Schwarzschilds Original Manifold and Its
Singularity, in General Relativity Research Trends, edited by A Reiner (Nova Science
Publishers, NY, 2006). Also arXiv: gr-qc/0406090 v2.
[20] S. Antoci and D-E Liebscher, The Topology of Schwarzschilds Solution and the
Kruskal Metric, arXiv:gr-qc/0308005 v3 (2005).
[21] S. Antoci and D-E Liebscher, Interpreting Solutions with Nontrivial Killing Groups
in General Relativity, arXiv: gr-qc/1007.4997v1 (2010).
[22] M. A. H. MacCallum, Gen. Relativ. Gravit. 38, 1887 (2006). Also arXiv:grqc/0608033v1 (2006).
[23] C. Corda, A Clarification on the Debate on the Original Schwarzschild Solution ,
EJTP, 8, 65 (2011).
[24] A. Kosowsky, private communication (2013).
[25] P.P. Fiziev, Gravitational Field of Massive Point Particle in General Relativity,
arXiv:gr-qc/0306088v12 (2003)
[26] R.E. Salvino and R.D. Puff, The Combridge-Janne Solution and the g22 Metric Function, preprint (2013). Available on academia.edu.
[27] H. A. Buchdahl. Isotropic Coordinates and Schwarzschild Metric, Int. J. Theor.
Phys., 24 731 (1985).
[28] A.S. Eddington, Mathematical Theory of Relativity (University Press, Cambridge,
1923).

30

Vous aimerez peut-être aussi