On the universality of free fall, the equivalence principle, and the

gravitational redshift
A. M. Nobili, D. M. Lucchesi, M. T. Crosta, M. Shao, S. G. Turyshev et al.

Citation: Am. J. Phys. 81, 527 (2013); doi: 10.1119/1.4798583

Related Articles
Microreviews by the book review editor: Einstein's Physics: Atoms, Quanta, and Relativity — Derived,
Explained, and Appraised: Ta-Pei Cheng
Phys. Teach. 51, 383 (2013)
A General Relativity Workbook.
Am. J. Phys. 81, 317 (2013)
Spatial geometry of the rotating disk and its non-rotating counterpart
Am. J. Phys. 80, 772 (2012)
The Standard Model appearing in Minute Physics video vignettes on YouTube
[Link]/watch?v=HVO0HgMi6Lc
Phys. Teach. 50, 253 (2012)
Embeddings and time evolution of the Schwarzschild wormhole
Am. J. Phys. 80, 203 (2012)

Additional information on Am. J. Phys.

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
On the universality of free fall, the equivalence principle,
and the gravitational redshift
A. M. Nobili
Department of Physics “E. Fermi,” University of Pisa, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy and
INFN-Istituto Nazionale di Fisica Nucleare, Sezione di Pisa, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy
D. M. Lucchesi
INFN-Istituto Nazionale di Fisica Nucleare, Sezione di Pisa, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy and
INAF-IAPS-Istituto di Astrofisica e Planetologia Spaziali, Via Fosso del Cavaliere 100, 00133 Roma, Italy
M. T. Crosta
INAF-Osservatorio Astrofisico di Torino, Pino Torinese, Torino, Italy
M. Shao and S. G. Turyshev
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Dr., Pasadena,
California 91109
R. Peron
INAF-IAPS-Istituto di Astrofisica e Planetologia Spaziali, Via Fosso del Cavaliere 100, 00133 Roma, Italy
G. Catastini and A. Anselmi
Thales Alenia Space Italia, Strada Antica di Collegno 253, 10146 Torino, Italy
G. Zavattini
Department of Physics, University of Ferrara, Via Saragat 1, 44122 Ferrara, Italy and
INFN-Istituto Nazionale di Fisica Nucleare, Sezione di Ferrara, Via Saragat 1, 44122 Ferrara, Italy
(Received 10 August 2012; accepted 15 March 2013)
Through the contributions of Galileo, Newton, and Einstein, we recall the universality of free fall
(UFF), the weak equivalence principle (WEP), and the strong equivalence principle (SEP), in order
to stress that general relativity requires all test masses to be equally accelerated in a gravitational
field; that is, it requires UFF and WEP to hold. The possibility of testing this crucial fact with null,
highly sensitive experiments makes these the most powerful tests of the theory. Following Schiff,
we derive the gravitational redshift from the WEP and special relativity and show that, as long as
clocks are affected by a gravitating body like normal matter, measurement of the redshift is a test
of UFF/WEP but cannot compete with direct null tests. A new measurement of the gravitational
redshift based on free-falling cold atoms and an absolute gravimeter is not competitive either.
Finally, we compare UFF/WEP experiments using macroscopic masses as test bodies in one case
and cold atoms in the other. We conclude that there is no difference in the nature of the test and
that the merit of any such experiment rests on the accuracy it can achieve and on the physical
differences between the elements it can test, macroscopic proof masses being superior in both
respects. VC 2013 American Association of Physics Teachers.
[[Link]

I. INTRODUCTION relativity and its foundation, the UFF, were few and required
measuring tiny effects to excruciating precision. General rel-
The universality of free fall (UFF)—that the acceleration ativity came to be regarded as a theory that was beautiful but
imparted to a body by a gravitational field is independent of uninteresting to experimental physicists. Things began to
the nature of the body—was established by Galileo and change in the late 1950s, at the dawn of the space age, when
Newton as an experimental fact, within the limits of the physicists realized that the means to test GR were becoming
experiments of their time. If it holds, inertial and gravitational available in the laboratory and, even more promising, in
mass are equivalent; this is the weak equivalence principle space.
(WEP). Newton made this “principle of equivalence” the ba- In 1958, a few months after the launch of the first artificial
sis of classical mechanics. More than two hundred years later, satellite by the Soviet Union, a committee appointed by U.S.
Einstein extended it to the invariance of physical laws in non- President Eisenhower—whose advice led to the establish-
rotating laboratories freely falling in a uniform gravitational ment of NASA the same year—wrote a famous pamphlet2 in
field, making it the foundation of his theory of general relativ- which physics was first in the list of scientific objectives soon
ity (GR). to be addressed in space. They proposed a space experiment
Because UFF is an experimental fact, it can be disproved to measure the gravitational redshift, which they regarded as
by experiment. For Einstein, the clever torsion balance a crucial test of GR (Ref. 2, p. 7): “Physicists are anxious to
experiments carried out by E€otv€os in Hungary at the turn of run one crucial and fairly simple gravity experiment as soon
the 20th century1 were proof enough. And there the matter as possible. This experiment will test an important prediction
stood for about fifty years. Experiments testing general made by Einstein’s general theory of relativity, namely that a

527 Am. J. Phys. 81 (7), July 2013 [Link] C 2013 American Association of Physics Teachers
V 527

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
clock will run faster as the gravitational field around it is Galileo was 74, almost blind, and under house arrest by order
reduced.” of the Pope, but he had made these experiments in the early
In January 1960, Leonard Schiff3 showed that, as long as a 1600s.11 The accuracy of the test was12 about 103 .
gravity field affects clocks and normal matter alike, the gravi- In 1687, in the opening paragraph of the Principia,13
tational redshift can be derived from WEP and special relativ- Newton wrote: “This quantity that I mean hereafter under
ity and as such it is not a crucial test of GR. Measurements of the name of…mass…is known by the weight…for it is pro-
the gravitational redshift are therefore tests of the underlying portional to the weight as I have found by experiments on
WEP, but he argued that even with the best atomic clocks of pendulums, very accurately made….” If inertial and gravita-
the time they could not compete with the torsion balance tests tional mass mi and mg are the same for all test bodies regard-
reported by E€ otv€
os half a century earlier. less of their mass and composition, the equations of motion
Recently, a proposal has been made for a new space mis- under the gravitational attraction of a source mass M (e.g.,
sion devoted to measuring the gravitational redshift and the Earth, assumed for simplicity to be spherically symmet-
testing WEP with cold atoms.4 A much better ground mea- ric) state that they all fall with the same acceleration:
surement of the gravitational redshift has been reported,
based on free falling cold atoms.5 A strong scientific debate GMmg
is ongoing and this has motivated the present work. r€ ¼ 
mi~ r ; so mi ¼ mg implies that
~
r3
This paper is organized as follows. In Sec. II, we go back GM
to UFF and the equivalence principle of Galileo, Newton, r€ ¼  3 ~
~ r: (1)
r
and E€ otv€
os to stress the advantages of null experiments. In
Sec. III, we recall how UFF and Newton’s equivalence prin- If inertial and gravitational mass are equivalent, UFF holds;
ciple (the “weak equivalence principle”) led Einstein to the should experiments invalidate UFF, they would invalidate
strong equivalence principle (or “Einstein equivalence the equivalence of inertial and gravitational mass as well.
principle”) and to the formulation of general relativity, so This was the “equivalence principle” from 1687 until
that a violation of UFF/WEP would either require that GR be 1907. Note that Eq. (1) holds for any position vector ~ r , in the
amended or call for a new force of nature. This has led to the vicinity of the source body as well as very far away from it,
quest for UFF/WEP null experiments that are as accurate e.g., it applies also to test bodies on Earth falling in the grav-
and precise as possible. itational field of the Sun (as tested by Dicke and his students
In Sec. IV, following Schiff,3 we derive the gravitational in the early 1960s14), or in the case of the Earth and the
redshift from the WEP and special relativity and show that— Moon falling towards the Sun (as tested with lunar laser
as long as clocks are affected by a gravitating body like nor- ranging15).
mal matter—its measurement is a test of UFF/WEP but it is E€otv€os and collaborators1 first coupled the test masses by
by far less competitive than direct null tests. A suggested suspending them on a very sensitive torsion balance, and
stronger deviation from WEP of clocks as compared to ordi- were able to test UFF in the field of the Earth to about 108 .
nary bodies6 appears to be beyond the reach of current and Dicke’s torsion balance experiment was the first UFF test in
planned experiments. The best, so-far controversial, measure- the field of the Sun (to ’ 1011 ), followed by Braginsky and
ment of the gravitational redshift,5 based on free falling cold Panov16 (to ’ 1012 ). More recent experiments with rotating
atoms in combination with a nearby absolute gravimeter, is a torsion balances have tested UFF both in the field of the
test of the WEP. As such it is in perfect agreement with the Sun17 and in the field of the Earth18 yielding the best limits
original experimental result,7 but it is in no way competitive to date (see Ref. 19, Table 3): UFF is confirmed to about
with UFF/WEP null tests. In this analysis we frequently step 1012 in the field of the Sun and to about 1013 in the field of
into the “Schiff conjecture” as formulated in 1973 by Thorne the Earth.
et al.8 In consideration of the “vigorous argument” between It is worth stressing that UFF experiments can reach high
Schiff and Thorne on this issue, we trace the conjecture back accuracy because they can be performed as null experiments.
to Schiff’s original statement in Ref. 3 and report it. We also The physical quantity of interest in UFF experiments is the
report the results of the best experiment to date that has com- relative acceleration Da ¼ a1  a2 of the free falling proof
pared the effect of a gravitating body (the Sun) on the rate of masses, from which the dimensionless E€otv€os parameter
clocks of different internal structure and in different locations
as the solar potential changes over the year.9 Da
Finally, in Sec. V, we compare UFF/WEP tests using mac- g (2)
a
roscopic proof masses versus cold atoms to show that,
although the experiments are completely different, there is is obtained (here a ¼ ða1 þ a2 Þ=2 is the average free fall
no difference in the nature of the tests and one should pursue acceleration of the masses in the gravity field of the source
the most promising ones, both in terms of sensitivity and in body). The g parameter quantifies a deviation from UFF. If
terms of differences in the physical properties of the atoms UFF holds, Da ¼ 0 and g ¼ 0; for a given value of a, the
being tested. smaller the differential acceleration measured, the smaller
the value of g, the more accurate the test.
II. UNIVERSALITY OF FREE FALL AND THE If the experiment is designed to measure the differential
EQUIVALENCE PRINCIPLE FROM GALILEO TO acceleration between the test masses, the experiment signal
NEWTON should be zero in the absence of UFF violation (after classi-
cal differential effects have been reduced to below the tar-
In the Discorsi10 (pp. 128–129; pp. 84–85 of the English get). In such null experiments no precise theoretical
edition) Galileo describes his tests of the universality of free prediction must be made which the measured signal should
fall (UFF) made with two pendulums of different composi- be compared to in order to obtain the physical quantity of
tion. The book was published in Leiden in 1638 when interest.

528 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 528

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
 Galileo’s UFF test with pendulums was most probably the the roof of a house, there exists for him during his
first example of an “almost” null experiment. He measured fall no gravitational field—at least in his
for how long the masses of the two pendulums—which he immediate vicinity.
had released as equally as he could—would keep in step
Einstein recalled the same facts in a speech given at
with each other. E€ otv€os arranged the masses on a torsion bal-
Kyoto University in 1922 entitled “How I created the theory
ance and measured the deflection angle of the suspension
of relativity”:23
fiber. In the absence of UFF violation there should be no dif-
ferential effect, hence no deflection.
While I was writing this [i.e., a summary essay on
special relativity], I came to realize that all the
III. EINSTEIN’S “HAPPIEST THOUGHT” AND THE natural laws except the law of gravity could be
STRONG EQUIVALENCE PRINCIPLE discussed within the framework of the special
theory of relativity. I wanted to find out the reason
In 1907, referring to his work “On the electrodynamics of
for this, but I could not attain this goal easily.
moving bodies” published two years earlier20 in which he
presented what is known as the special theory of relativity,
Einstein wrote as follows (Ref. 21, Ch. V, “Principle of rela- …The breakthrough came suddenly one day. I
tivity and gravitation,” Sec. 17, “Accelerated reference sys- was sitting on a chair in my patent office in Bern.
tem and gravitational field”): Suddenly a thought struck me: If a man falls
freely, he would not feel his weight. I was taken
Until now we have applied the principle of aback. This simple thought experiment made a
relativity—i.e., the assumption that the laws of big impression on me. This led me to the theory
nature are independent of the state of motion of gravity. I continued my thought: A falling man
of the reference system—only to nonaccelerated is accelerated. Then what he feels and judges is
reference systems. Is it conceivable that the happening in the accelerated frame of reference. I
principle of relativity holds also for systems that decided to extend the theory of relativity to the
are accelerated relative to each other? reference frame with acceleration. I felt that in so
doing I could solve the problem of gravity at the
same time.
…We consider two systems of motion R1 and R2 .
Suppose R1 is accelerated in the direction of its X
axis, and c is the magnitude (constant in time) of …It took me eight more years until I finally
this acceleration. R2 is at rest, but situated in a obtained the complete solution.
homogeneous gravitational field, which imparts to Thus, the well tested UFF leads to the statement that a
all objects an acceleration c in the direction of frame at rest in a (uniform) gravitational field and a uni-
the X axis. formly accelerated frame free from gravitational fields (with
uniform linear acceleration equal and opposite to the acceler-
As far as we know, the physical laws with respect ation produced in the gravitational frame) are equivalent for
to R1 do not differ from those with respect to R2 ; all physical processes.
this derives from the fact that all bodies are This is a crucial leap from Newton’s equivalence princi-
accelerated alike in the gravitational field. We ple, as stated in Sec. II (now referred to as the weak equiva-
have therefore no reason to suppose in the present lence principle WEP), to the strong equivalence principle
state of our experience that the systems R1 and R2 SEP (also referred to as the Einstein equivalence principle
differ in any way, and will therefore assume in EEP24). It is apparent that should experiments invalidate
what follows the complete physical equivalence of UFF (and the WEP), they would invalidate the SEP as well.
the gravitational field and the corresponding In his influential Les Houches lectures, given in 1963 and
acceleration of the reference system. published in 1964,25 Robert Dicke stated the strong equiva-
lence principle in such a precise, clear, and simple manner
In a 1919 manuscript (Ref. 22, p. 364) Einstein wrote:
that it is worth using his definition (Ref. 25, p. 4):
When, in the year 1907, I was working on a The strong equivalence principle might be defined
summary essay concerning the special theory of as the assumption that in a freely falling, non-
relativity for the Jahrbuch fuer Radioaktivitaet und rotating, laboratory the local laws of physics take
Elektronik, I had to try to modify Newton’s theory on some standard form, including a standard nu-
of gravitation in such a way that it would fit into merical content, independent of the position of the
the theory. Attempts in this direction showed the laboratory in space and time. It is of course implicit
possibility of carrying out this enterprise, but they in this statement that the effects of gradients in the
did not satisfy me because they had to be gravitational field strength are negligibly small,
supported by hypotheses without physical basis. At i.e., tidal interaction effects are negligible.
that point, there came to me the happiest thought
of my life, in the following form: A few sentences below, Dicke adds (Ref. 25, p. 5): “It is
well known that this interpretation of the equivalence princi-
Just as is the case with the electric field produced ple, plus the assumption of general covariance is most of
by electromagnetic induction, the gravitational what is needed to generate Einstein’s general relativity.”
field has similarly only a relative existence. For if The route that led Einstein from SEP to the field equations
one considers an observer in free fall, e.g., from of gravitation26 and the general theory of relativity27

529 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 529

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
required moving beyond Euclidean geometry. If all acceler-
ated frames are equivalent, then Euclidean geometry cannot
hold in all of them: gravity is locally replaced by a uniformly
accelerated system, hence globally there cannot be just one
such system. On the other hand, as Einstein put it:
“Describing the physical laws without reference to geometry
is similar to describing our thought without words. We need
words in order to express ourselves.”23 Luckily, the required
non-Euclidean geometry and related mathematics were
available, after Gauss’ original intuitions, thanks to the work
of Riemann and other mathematicians.
In summary, GR is founded on SEP, which requires UFF
to hold. Should experiments invalidate UFF, either GR must Fig. 1. Illustration from Ref. 3 where the caption reads: “(a) Two
be amended because it is not a fully correct theory of gravity identically-constructed clocks, A and B, are at rest in a gravitational field.
or else we are in the presence of a new, so far unknown, (b) The gravitating body is replaced by an upward acceleration g of clocks A
and B, and a stationary clock C is introduced to compare their periods.”
physical interaction. UFF experiments test a fundamental
physical principle; moreover, they can reach extremely high
accuracy because they can be performed as null experiments. uniform field of the gravitating body, can then be replaced
This is why they are extremely powerful probes of funda- by the same clocks accelerated with equal and opposite
mental physics, worth improving whenever possible. acceleration in empty space, their equations of motion being
In 1916 the best experimental tests of UFF/WEP were the same in the two cases. WEP is all that is needed to make
those by E€ otv€os1—initiated in 1889—and Einstein explicitly the two configurations equivalent to each other. Schiff chose
recognized their relevance (Ref. 27, Sec. 2): to reason in configuration (b), as follows. An observer in an
inertial reference frame has three identical clocks A, B, and
C. When they are at rest they all tick with the same period T.
…This view is made possible for us by the
Then, clocks A and B are accelerated with the same constant
teaching of experience as to the existence of a field
acceleration g, clock A being the leading one, while clock C
of force, namely the gravitational field, which
remains at rest and is used for comparison.
possesses the remarkable property of imparting the
As A and B are accelerated upward with the same acceler-
same acceleration to all bodies.*
ation g, clock A will pass by clock C having velocity vA > 0,
*E€otv€
os has proved experimentally that the
and therefore it is seen by C as ticking with a slightly longer
gravitational field has this property in great
period TA with respect to it. Instead, clock B will pass by
accuracy. [Footnote in the original.]
C—which is still ticking with period T—having a somewhat
higher velocity vB > vA , and therefore ticking with an even
IV. THE GRAVITATIONAL FREQUENCY SHIFT longer period TB. This is because of the time dilation of spe-
In 1960 Schiff,3 by applying to clocks the WEP as tested cial relativity, which implies
for ordinary bodies, pointed out—and demonstrated—that T T
the gravitational redshift can be derived solely from the TA ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi; TB ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi (3)
weak equivalence principle and special relativity, as Einstein 1  v2A =c2 1  v2B =c2
did in 1911 (see Ref. 28, pp. 101–103). While Einstein
referred to sources of radiation, Schiff—similarly to Tolman (with c the speed of light). Combining these two equations
(Ref. 29, p. 192)—used clocks. gives
Schiff derived the gravitational redshift in a simple and pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
straightforward manner, limiting the assumptions required to 1  v2A =c2
TB ¼ TA pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi: (4)
the very minimum. Concerning experiments to measure it, he 1  v2B =c2
concluded: “Terrestrial or satellite experiments that would go
beyond supplying corroborative evidence for the equivalence By expanding Eq. (4) in power series of v2A =c2 and v2B =c2 , we
principle and special relativity would be very difficult to per- have
form, and would, for example, require a frequency standard    
with an accuracy somewhat better than 1 part in 1018 .” 1 v2B  v2A gh
TB ’ TA 1 þ ’ TA 1 þ 2 ; (5)
We re-derive Schiff’s result and show that nowadays 2 c2 c
measurements of the gravitational redshift have no chance of
matching the current level of ground based tests of UFF/ where we have used v2B ¼ v2A þ 2gh because the clocks are
WEP, and even less so the more accurate ones to be per- moving with uniform acceleration g with respect to the sta-
formed in space. tionary frame of clock C, which is at rest in a gravitation-free
As long as WEP holds for clocks and normal matter alike, region.
the configurations shown as (a) and (b) in Fig. 1 (taken from The general case of a nonuniform gravitational field (i.e.,
Ref. 3) are locally equivalent. “Locally” equivalent means when h is no longer negligible compared to the distance from
that they are equivalent as long as the two clocks A and B the center of mass of the gravitating body) is not substantially
are separated (along the lines of force) by a distance h much different. As Schiff argues, now reasoning in configuration
smaller than their distance from the center of mass of the (a), gh can be replaced by the difference DU in the gravita-
gravitating body (e.g., the Earth). In this case tidal effects tional potential between the positions of clocks A and B. This
between them are negligible and the gravity field they are requires one to perform not just one inter-comparison of
immersed in is uniform. The clocks, both at rest in the clocks but a series of such inter-comparisons between a

530 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 530

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
number of (identical) clocks arranged in such a way that the Newton, namely, their inertial and gravitational masses Mi,
gravitational field is nearly uniform from one inter- Mg and mi, mg may not be identical. In this case, we write
comparison to the next. WEP is valid at all distances from the
source body, so there will always be a configuration of type Mg ¼ Mi ð1 þ ge Þ; mg ¼ mi ð1 þ gc Þ; (9)
(b) where Schiff’s arguments hold. Therefore, in a nonuni-
form gravitational field and as long as second-order terms can where ge and gc may differ from zero, quantifying the devia-
be neglected, the periods will be related by tion from equivalence for the source body and the clock,
  respectively. In the presence of a possible violation of the
DU weak equivalence principle, each clock will be affected by
TB ¼ TA 1 þ 2 (6) the gravitating body according to the equation (see Ref. 33,
c
Sec. 2)
with
GMi ð1 þ ge Þmi ð1 þ gc Þ
r€ ¼ 
mi~ ~
r; (10)
GM GM r3
DU ¼  þ > 0; (7)
rA rB
and, since UFF/WEP experiments have confirmed the equiv-
where M is the mass of the gravitating source body as alence of gravitational and inertial mass to very small values
depicted in Fig. 1(a) (assumed for simplicity to have spheri- of g, we can write
cal symmetry), and rA and rB are the radial distances of
GMi ½1 þ g þ Oðg2 Þ
clocks A and B, respectively, from its center of mass. r€ ¼ 
~ ~
r; (11)
Equation (6) expresses the gravitational redshift D= ¼ r3
ð A   B Þ= B > 0 ( in the denominator indicates the fre-
quency of clock B) stating that the clock at lower altitude with g ¼ ge þ gc . With a separation distance much smaller
(e.g., on the surface of the Earth) is red-shifted if compared than the distance from the center-of-mass of the source body,
to an identical one at higher altitude (e.g., on board of a the clocks are subject to a uniform acceleration of magnitude
spacecraft); that is, GMi
g0 ¼ g½1 þ g þ Oðg2 Þ; g¼ : (12)
D DU r2
¼ 2 : (8)
 c If they are separated by a distance comparable to the distance
It is important to stress that this result has been obtained from the source body, they now experience a gravitational
using neither the conservation of energy nor the mass-energy potential difference,
equivalence.30 It refers only to identically constructed clocks
located at different distances from the center of mass of a DU0 ¼ DU½1 þ g þ Oðg2 Þ: (13)
gravitating source body along the lines of force. All that is
Equation (8) for the gravitational redshift is correct only as
required is that the clocks obey the WEP like ordinary bodies
long as g ¼ 0, which implies Mg ¼ Mi  M, mg ¼ mi  m,
[which makes configurations (a) and (b) of Fig. 1 equivalent
g0 ¼ g, and DU0 ¼ DU. If g ¼ ge þ gc 6¼ 0, since the clocks
to each other] and the special theory of relativity20 [which
are identical, they are still subject to the same acceleration
provides the time dilation in Eq. (3)].
from the gravitating body, hence configurations (a) and (b) of
By Ockham’s razor, this demonstration is preferable to
Fig. 1 are still equivalent to each other and Schiff’s method is
others that require more assumptions.
still correct. However, as shown by Eq. (12), the clocks are
In the experiment we discuss below,9 the clocks are in free
subject to an acceleration slightly different than in the case
fall relative to the source body, while in GP-A (Gravity
g ¼ 0 and the gravitational frequency shift involves, to first-
Probe A),31 one clock is in free fall and the other is at rest on
order in g, DU0 ¼ DUð1 þ gÞ. We therefore have
the surface of the Earth. As pointed out by Pegna,32 these
facts should not generate confusion as to how the gravita-     
tional redshift is derived: the clocks of Fig. 1(a) are at rest D DU DU
¼ 2 1þgþO 2 ; (14)
relative to the gravitating body, not in free-fall like in an  g c c
“Einstein elevator.” Inside a freely falling elevator, as long
as the field is uniform (locally), they would be subject to where the order of magnitude of the contribution from gravi-
zero total force, which is equivalent to being inside an eleva- tational redshift to second-order has been introduced since it
tor at rest in empty space (or moving with uniform velocity), may be needed in comparison with g. For the purpose of the
in which case there would be no frequency shift. In Ref. 9, comparison, we consider an ideal experiment in which the
clocks are in free fall together with the Earth relative to the clocks are at rest relative to each other and the gravitating
Sun, and the frequency shift occurs due to the variation of body is spherically symmetric and non-rotating; in this case
the field with the eccentricity of the Earth’s orbit. In GP-A,31 all contributions of order 1=c3 are zero.34 As for terms of
for a clock in free fall near Earth (after separation of the last order 1=c4 , a detailed derivation performed in the PPN (para-
rocket stage), as the gravitational potential felt by it changes, metrized post-Newtonian) framework can be found in
its frequency relative to the clock at rest is shifted. Ref. 35—including the contribution due to the rotation of the
We now assume that special relativity is fully valid (as gravitating body. However, only their order of magnitude is
tested by high-energy particle physics) while UFF/WEP relevant here.
might be violated at some level for clocks and ordinary mat- Equation (14) shows that the measurement of the gravita-
ter alike. The masses of the source body and the clocks are tional redshift is affected by a WEP violation even if it
now allowed not to obey the weak equivalence principle of involves a single set of clocks, all with the same composition

531 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 531

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
(same internal structure). If, in addition, the frequency shift where the first term ðus  ue Þ=c2 , is the difference in gravi-
is measured with another set of identical clocks A0 , B0 , C0 tational potential between the clock on the spacecraft and the
that are different in composition from the set A, B, C, the clock on the ground divided by c2 (it is our DU=c2 ).
shift measured will be different if gc0 6¼ gc . Assuming that all Two more terms had to be taken into account in order to
other experimental conditions are equal, the difference in the predict the correct value of the expected first-order gravita-
frequency shifts measured with the two sets of clocks will tional redshift. The second term contains the square of the
depend only on the difference gc  gc0 , while ge is eliminated relative velocity j~ v s ~v e j between the spacecraft and the
and there is no dependence on the composition of the source Earth and it is the second-order Doppler effect (the first-
body. Similarly, in the case of UFF tests that measure the order Doppler effect was nicely canceled in the experiment).
differential acceleration of two proof masses of different The third term is the residual first-order Doppler shift due to
composition freely falling in the gravitational field of a the acceleration ~ a e of the ground station during the light trav-
source body, only the materials constituting the proof masses eling time between the spacecraft and the ground station,
are relevant (the experiments test g ¼ g2  g1 , subscripts 1 j~
r se =cj.
and 2 referring to the two proof masses), while the composi- As shown in Eq. (16), the measured value of ðD=ÞGPA
tion of the source body does not affect the measurement.33 agrees with the theoretical prediction to the level 1:4  104,
In summary, as long as clocks are affected by a gravita- which makes g as tested with UFF experiments (to 1012 at
tional field like normal matter, if the possibility of WEP vio- the time and to 1013 since 2008) totally inaccessible to GP-
lation is taken into account, the general expression for A. If the result had shown a discrepancy from the first-order
gravitational redshift is given by Eq. (14). This result shows prediction, the experiment itself should have been carefully
that a measurement of the gravitational redshift does also scrutinized. In this experiment the second-order term of the
test the weak equivalence principle and can therefore be redshift amounts to ’ 1:8  1019 and in the PPN framework it
compared with UFF experiments. contains the b and c parameters35 (b ¼ c ¼ 1 in general rela-
Limits on g are set by UFF experiments for the specific tivity). That is, to second order the WEP no longer accounts
materials employed, which are selected with the purpose of for the gravitational redshift and Schiff’s derivation of it is
maximizing the chance of violation. We regard g measured inapplicable, as he noted in his concluding remarks (Ref. 3, p.
by UFF as an upper limit for Eq. (14), meaning that if 343). However, this term is too small to be detected even with
a frequency standard of accuracy 1018 and it would not be
  detected by the STE-QUEST experiment4 either.
DU
g>O ; (15) The GP-A measurement of the gravitational redshift
c2
shows clearly the difference with respect to a null experi-
and provided that D= is measured to somewhat better than ment. The measured frequency shift had to be compared
gDU=c2 , it follows that the gravitational redshift experiment with the sum of the three terms on the right-hand side of Eq.
would test WEP better than UFF experiments. In a real space (16), whose values depend on various physical quantities,
experiment, because of the motion of the clocks, additional some of which had to be measured during the experiment
terms would appear in Eq. (14), larger than the one to order itself. It is only by comparing the theoretical prediction and
1=c4 . These terms are of order 1=c2 [see Eq. (16)] as well the measured shift that the authors could establish the ratio
as34 1=c3 , and they all need to be adequately modeled. ½1 þ ð2:5 6 70Þ  106 . No wonder that it took them four
Schiff’s paper appeared in January 1960. At that time, the years to publish the results of an experiment that lasted only
best UFF test had been performed by the E€otv€os group sev- about two hours! The advantage of a null UFF experiment is
eral decades earlier,1 to g ’ 108 . The space age had just apparent and we are not surprised to find out that UFF
begun and the GP-A mission proposal, to launch an atomic experiments are superior.
clock to high altitude and compare it with an identical clock Moreover, as clocks become more accurate, the theoretical
on ground, was already under discussion. In such a situation modeling of all the contributing terms and the measurement
(DU=c2 ’ 4:3  1010 ; this value would slightly increase to to a comparable accuracy of all the physical parameters
6  1010 with STE-QUEST4), inequality (15) holds and involved in them become more and more difficult.34,35
Schiff estimated that a frequency standard with an accuracy The accuracy of frequency standards has considerably
somewhat better than one part in 1018 was required for a improved since GP-A. However, since 2008 UFF has been
measurement of the gravitational redshift to improve on the tested to g ’ 1013 in the field of the Earth,18 hence the g
UFF tests by E€ otv€os. Such a good frequency standard was term in Eq. (14) is much smaller than the second-order fre-
far beyond reach at the time. Hence Schiff concluded that a quency shift. Even pushing such measurement to second
space experiment to measure the gravitational redshift would order would—in the words of Schiff—only provide corrobo-
have limited significance. rative evidence for the weak equivalence principle. The cur-
GP-A was launched by NASA 16 years later, in 1976. rent values of the b and c parameters, which are involved in
Two identical hydrogen maser clocks were used to measure the second-order term, would not be improved either.
the gravitational redshift, one on the ground and the other on In May 1960 Schild36 analyzed Schiff’s derivation of the
board a spacecraft launched to 10,000 km altitude. The gravitational redshift, finding that it is correct (though only
result31 was reported in 1980: to first order, as Schiff himself mentioned). He also showed
that, unlike the gravitational redshift, light bending cannot
  be derived purely from the equivalence principle and special
D
¼ ½1 þ ð2:5 6 70Þ  106  relativity as Schiff did in the same paper. In 1968 Rindler37
 GP-A too showed that Schiff’s procedure is legitimate for clocks
!
us  ue j~ v s ~v e j2 ~r se  ~
ae but spurious for rods.
   ; (16) Was Schiff correct in applying to clocks the WEP as tested
c2 c2 c2
for ordinary bodies?

532 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 532

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
 In 1975, a year before the launch of GP-A, Nordtvedt6 As we understand it, Schiff is arguing that the rates of
investigated the gravitational redshift using in addition the clocks must be affected by a gravitational field as expected
conservation of energy. He concluded that WEP violation for normal matter, i.e., all in the same way regardless of their
might affect the clocks used to measure the gravitational red- internal structure.
shift more strongly than it would affect the ordinary proof Many years later, Schiff’s statement has been reformu-
masses used to test UFF, thus invalidating Schiff’s derivation lated by Thorne et al.8 as follows, and ever since named the
of the gravitational redshift. The “amplification factor” would Schiff conjecture: “Any complete and self-consistent gravita-
depend upon the specific energy being rearranged in the phys- tion theory that obeys WEP must also, unavoidably, obey
ical process taking place in the clocks to generate a frequency EEP.” (Einstein equivalence principle24 is another way of re-
standard. He quantified the amplification factor as the ratio of ferring to the strong equivalence principle.25)
the atom rest mass to the mass-energy involved in the fre- Though the Schiff conjecture is very frequently quoted
quency generation process. This amounts to saying that the in the literature, we prefer to refer to Schiff’s original
closer the mass-energy involved in frequency generation is to statement (and to his paper) also in consideration of the
the rest mass-energy of the atom, the closer the clock is to fact that a “vigorous argument” took place on this issue
normal matter. For the hydrogen maser clock, assuming that between Schiff and Thorne at the 1970 Caltech-JPL con-
all the mass-energy involved in frequency generation violates ference on experimental tests of gravitational theories, af-
WEP, Nordtvedt estimated that it could “amplify” g measured ter which Schiff had no chance to go back to the subject in
by UFF experiments by about four orders of magnitude. With writing because he died just two months later (Ref. 8,
g ’ 1012 at the time, even a violation four orders of magni- p. 3577).
tude larger was still too small to be measured by far. An experiment testing the effect of a gravitational field on
Nordtvedt regarded this kind of contribution to gravitational the rates of different clocks has been performed by Ashby
redshift as a test of energy conservation. et al.9 The rate of any clock on the ground is affected by the
Back in 1960, on the next page after Schiff’s paper, gravity field of the Sun. Because the solar potential varies
Dicke38 argued against it on the grounds that the results of over the year due to the eccentricity of the Earth’s orbit, the
UFF tests cannot be extended to clocks. The dispute appears rate of the clock undergoes an annual variation due to the
to have been affected by the ongoing discussion about gravitational redshift from the Sun. Should two nearby
NASA funding a space mission (GP-A) devoted to testing clocks of different internal structure be affected differently
the gravitational redshift, and the two papers next to each by the gravity field of the Sun, a difference will appear in
other are very frequently quoted together in the literature. their annual frequency shifts. Clocks farther apart on the sur-
However, it is worth recalling that a few years later, in 1964, face of the Earth can also be compared. Over a timespan of
Dicke agreed with Schiff. He wrote (Ref. 25, p. 5): “The red seven years, Ashby et al.9 compared the frequencies of four
shift can be obtained from the null result of the E€otv€os hydrogen masers at NIST (USA) with one cesium fountain
experiment, mass energy equivalence, and the conservation clock in the same lab, and also with three more cesium foun-
of energy in a static gravitational field and static coordinate tain clocks in Europe (in Germany, France, and Italy). The
system.” He derived the gravitational redshift from these result is that the annual variation of the gravitational poten-
assumptions, the mass-energy equivalence being confirmed tial of the Sun produces on all pairs of clocks the same fre-
by the Hughes39 and Drever40 experiments. A few pages quency shift to 1:4  106 , despite their different structure
later, while discussing the three famous tests of GR, on the and also different location on the surface of the Earth. A sim-
gravitational redshift Dicke wrote (Ref. 25, p. 25): “While ilar experiment in space4 must take into account additional
this experiment may not be the most important of relativity terms (to order 1=c2 and 1=c3 ) due to the motion of the
experiments, it is interesting, and I should like to discuss clocks. Should a discrepancy be found, its interpretation
briefly the experiment of one of my students, Brault, on the would be very difficult.
redshift of solar lines.” In 2010 a measurement of the gravitational redshift four
Schiff commented on Dicke’s arguments of 1960 with a orders of magnitude better than GP-A was reported.5 It was
note added in proof to his paper (Ref. 3, p. 343), which reads: obtained by re-interpreting the data of a 1999 experiment7
that measured the absolute value of the local gravitational
The E€ otv€
os experiments show with considerable acceleration g using cold cesium atoms and atom interferom-
accuracy that the gravitational and inertial mass of etry. In addition, the experiment performed a test of UFF by
normal matter are equal. This means that the comparing this value with the value of g as measured by an
ground state Hamiltonian for this matter appears absolute gravimeter located nearby in the lab in which a laser
equally in the inertial mass and in the interaction of interferometer monitors the motion of a freely falling corner-
this mass with a gravitational field. It would be cube retroreflector (after correcting for known systematics
quite remarkable if this could occur without the like tides, polar motion, and others).
entire Hamiltonian being involved in the same way, The frequency affected by gravitational redshift in this
in which case a clock composed of atoms, whose case would be the Compton frequency xC ¼ mh=c2 , with m
motions are determined by this Hamiltonian would the rest mass of the free falling cold atom (cesium) and h the
have its rate affected in the expected manner by a reduced Planck constant. The gravitational redshift is recov-
gravitational field. Nevertheless, as stated in the ered from the atom interferometry signal—which contains
foregoing, I believe that a direct demonstration that the local gravitational acceleration g—once g has been meas-
the equivalence principle is valid for clocks would ured with the absolute gravimeter. The method has been
be useful. On the other hand, it is evident that questioned, and a strong dispute is ongoing.41–46 Here, we
experiments of this type could not verify any wish to stress the following facts. In this measurement the
feature of general relativity theory other than the expected first-order frequency shift is extremely small
first-order change in the time scale. (because h ’ 103 m or less); nevertheless, the accuracy is

533 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 533

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
remarkable (7  109 ), better than the goal of the STE- k-shell electrons have a 15 per cent increase in
QUEST proposal in space.4 their mass as a result of their relativistic velocities.
As stated by the authors44 and agreed upon by their oppo- Third, the electromagnetic negative contribution to
nents,45 in this case the mass-energy involved is the full the binding energy of the nucleus varies as Z2 and
mass-energy of the freely falling atom, just as in the previous represents 1/2 per cent of the total mass of a gold
WEP test,7 which is indeed the same (and only) experiment atom, whereas it is negligible in aluminum. In sim-
performed. Therefore Nordtvedt’s6 amplification factor ilar fashion, the virtual pair field around the gold
would be essentially unity. Both the corner-cube reflector nucleus would be expected to represent a far big-
and the cesium atoms must obey UFF (see also Sec. V), and ger contribution to the total energy than in the case
therefore a deviation from the predicted gravitational redshift of aluminum. Also, the virtual pion field, and other
would show a deviation from UFF/WEP according to Eq. virtual fields, would be expected to be different in
(14), the materials involved being cesium and glass. the two atoms. We would conclude that in most
The measured frequency shift is found to differ from the physical aspects gold and aluminum differ substan-
predicted gravitational redshift by5 b ¼ ð7 6 7Þ  109 (note tially from each other and that the equality of their
that the authors use the b symbol with no reference to the accelerations represents a very important condition
homonymous parameter in the PPN framework). to be satisfied by any theory of gravitation.
Eleven years earlier,7 with the same experimental appara-
tus and the same data, the authors stated: “We show that the Dicke is listing all known different physical properties of
macroscopic glass object used in this instrument [i.e., the an atom of gold as compared to an atom of aluminum. It
absolute gravimeter] falls with the same acceleration, to does not matter at all, in principle, how many of them are
within seven parts in 109 , as a quantum-mechanical caesium used in the experimental test.
atom.” After describing the absolute gravimeter and how its Experiments with macroscopic test bodies naturally have
run was used to compare with g as measured by the atom at their disposal a very large number of atoms, typically
inteferometer, they concluded: “A comparison with the value many times Avogadro’s number, while in cold atom experi-
of g we obtained in a two-day run shows a difference of ments 109 atoms would already be a very large number. The
ð7 6 7Þ p.p.b.” difference is enormous, but if the two experiments test the
As we can see, the 2010 and 1999 results agree perfectly same atoms, they should confirm or violate UFF/WEP just
with each other, but it is apparent that the accuracy of UFF the same because there is no difference in the mass-energy
null tests is totally inaccessible to this experiment (7  109 content of the atoms being tested. The way each experimen-
vs. g ’ 1013 already achieved18 and g ’ 1017 possible in talist “manufactures” his/her own “test masses,” the way he/
space47). she deals with them (by arranging them, manipulating them,
etc.), as well as the way their motion is read in order to
extract the required signal (i.e., their differential accelera-
V. TESTS OF THE UNIVERSALITY OF FREE FALL tion, which must be zero for UFF to hold), will obviously be
AND MASS–ENERGY CONTENT OF THE TEST completely different in the two cases. Although extremely
BODIES important for the outcome of the test and the sensitivity it
can achieve, these are technicalities.
The total mass-energy of a body can be expressed as the As pointed out by Pegna,48 unless evidence is provided
sum of many terms, corresponding to the energy of all that the preparation and manipulation of cold atoms in atom
the conceivable interactions and components: m ¼ Rk mk . interferometers alters the mass-energy content of the atoms
The dimensionless E€otv€os parameter g ¼ 2½ðmg =mi Þa by an amount accessible to the sensitivity of the test, there is
ðmg =mi Þb =½ðmg =mi Þa þ ðmg =mi Þb , which quantifies the no reason to expect a result different from a test with macro-
violation of equivalence for two bodies of composition a scopic bodies of the same composition.
and b, inertial mass mi, and gravitational mass mg, is then The fact that macroscopic masses and cold atoms fall with
generalized to the same acceleration in a gravitational field has been proved
to 7  109 (Ref. 7). The experiment compared the absolute
2½ðmg =mi Þak  ðmg =mi Þbk  value of the free fall gravitational acceleration of cesium
gk ¼ ; (17)
ðmg =mi Þak þ ðmg =mi Þbk cold atoms measured by atom interferometry with the abso-
lute value of the free fall gravitational acceleration of a
nearby macroscopic corner-cube reflector measured with
such that a nonzero value of gk would define the violation of laser interferometry. It is apparent that this is not a null
equivalence between the inertial and gravitational mass- experiment, which is definitely a disadvantage.
energy of the kth type. The rest mass would contribute to Since macroscopic test masses have been shown18 to obey
almost all the mass-energy content of the body and therefore UFF to 1013 (and this is certainly an upper limit), it follows
to almost all the measured Da and g  Da=a. that the experiment of Ref. 7 proves that cesium cold atoms
Dicke’s group used aluminum and gold.14 Dicke investi- and any macroscopic test mass fall with equal gravitational
gated their difference in great detail. In Ref. 25 (p. 4) he acceleration to 7  109 . Thus, in particular, cesium cold
wrote: atoms and a macroscopic body made of cesium would fall
…gold and aluminum differ from each other rather with equal gravitational acceleration to this level.
greatly in several important ways. First, the Through a purely theoretical analysis, Storey and
neutron to proton ratio is quite different in the two Cohen–Tannoudji demonstrated that in a uniform gravita-
elements, varying from 1.08 in aluminum to 1.5 in tional field the quantum propagator of any object is deter-
gold. Second, the electrons in aluminum move mined by the action along the classical path.50 Based on this
with non-relativistic velocities, but in gold the result, Unnikrishnan argued that the same is true for

534 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 534

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
accelerated objects in free space.51 Therefore, the outcome of Caltech, under a contract with NASA. Thanks are due to F.
any test of WEP performed in a classical setup will hold in a Pegoraro for helping to present this work in the context of
quantum context. the current debate.
It is in general very interesting to perform the same test
using completely different experiment concepts and experi- 1
R. V. E€ otv€os, D. Pekar, and E. Fekete, “Beitr€age zum Gesetze der
mental apparatus, because systematic errors will be com- Proportionalit€at von Tr€agheit und Gravit€at,” Ann. Phys. 373, 11–66
pletely different and possibly better understood. The (1922).
2
advantage is noticeable if different experiments have compa- The White House, “Introduction to outer space. An explanatory statement
rable accuracy. The weak equivalence principle has been prepared by the President’s Science Advisory Committee,” available online
at <[Link] 1–15 (1958).
tested with macroscopic test masses, of widely different 3
L. I. Schiff, “On experimental tests of the general theory of relativity,”
composition, to an accuracy of 1013 . Experiments with cold Am. J. Phys. 28, 340–343 (1960).
atoms have reached 107 (see Ref. 49); moreover, they have 4
ESA, “STE-QUEST: Space-time explorer and quantum equivalence prin-
used 85 Rb and 87 Rb, whose difference in composition by two ciple space test,” available online at <[Link]
neutrons limits the physical relevance of the test. 5
object/[Link]?fobjectid=49307>, 1–39 (2011).
The challenge is for a breakthrough by several orders of H. M€ uller, A. Peters, and S. Chu, “A precision measurement of the gravita-
tional redshift by the interference of matter waves,” Nature 463, 926–929
magnitude, which would explore a so-far unknown physical
(2010).
domain where chances for a major discovery are higher. The 6
K. Nordtvedt, “Quantitative relationship between clock gravitational ‘red-
figure of merit for the competing experiments is the accuracy shift’ violations and nonuniversality of free-fall rates in nonmetric theories
to which they can test UFF between materials as different as of gravity,” Phys. Rev. D 11, 245–247 (1975).
7
possible in their fundamental physical properties. At the A. Peters, K. Y. Chung, and S. Chu, “Measurement of gravitational accel-
present state of the art, such a breakthrough is more likely to eration by dropping atoms,” Nature 400, 849–852 (1999).
8
occur with macroscopic test masses than with cold atoms. K. S. Thorne, D. L. Lee, and A. P. Lightman, “Foundations for a theory of
gravitational theories,” Phys. Rev. D 7, 3563–3578 (1973).
9
N. Ashby, T. P. Heavner, S. R. Jefferts, T. E. Parker, A. G. Radnaev, and
Y. O. Dudin, “Testing local position invariance with four cesium-fountain
VI. CONCLUSIONS
primary frequency standards and four NIST hydrogen masers,” Phys. Rev.
In this paper, we have discussed experiments that test the Lett. 98, 070802-1-4 (2007).
10
G. Galilei, Discorsi e dimostrazioni matematiche intorno a due nuove sci-
universality of free all and the weak equivalence principle in enze attenenti alla mecanica & i movimenti locali, in Edizione Nazionale
comparison with experiments that measure the gravitational delle Opere di Galilei, Vol. VIII, p. 128, Barbera Ristampa del 1968,
redshift. Quoting from Einstein’s original writings, we have Firenze; Dialogues Concerning Two New Sciences (Dover, New York,
shown how UFF and the weak equivalence principle led him 1954).
11
to the strong equivalence principle (or Einstein equivalence D. Bramanti, G. Catastini, A. M. Nobili, E. Polacco, E. Rossi, and R.
principle) and to the formulation of general relativity, so that Vergara Caffarelli, “Galileo and the universality of free fall,” in
Proceedings of the STEP Symposium, Pisa, Italy 6-8 April 1993, ESA
a violation of UFF would either require that general relativity
WPP-115, p. 319.
be amended, or call for a new force of nature. Hence, the 12
F. Fuligni and V. Iafolla, “Galileo and the principle of equivalence,” in
need of testing UFF as accurately as possible. Proceedings of the STEP Symposium, Pisa Italy 6–8 April 1993, ESA
Following Schiff, we have concluded that a measurement WPP-115, pp. 104–109.
13
of the gravitational redshift is a test of the weak equivalence F. Cajori, Sir Isaac Newton’s Mathematical Principles of Natural
principle but one that—being an absolute measurement—can- Philosophy (University of California, Berkeley, 1934).
14
not compete with the direct null tests of the universality of P. G. Roll, R. Krotkov, and R. H. Dicke, “The equivalence of passive and
gravitational mass,” Ann. Phys. 26, 442–517 (1964).
free fall. Nordtvedt’s analysis, whereby clocks might violate 15
J. G. Williams, S. G. Turyshev, and D. H. Boggs, “Progress in lunar laser
the weak equivalence principle to a level much stronger than ranging tests of relativistic gravity,” Phys. Rev. Lett. 93, 261101-1–4
normal matter, implies that a competitive accuracy would still (2004).
16
be inaccessible to current and planned redshift experiments. V. B. Braginsky and V. I. Panov, “Verification of the equivalence of iner-
A recent measurement of the gravitational redshift, per- tial and gravitational mass,” Sov. Phys. JEPT 34, 463–466 (1972).
17
formed with data from a previous experiment that used free S. Baeßler, B. R. Heckel, E. G. Adelberger, J. H. Gundlach, U. Schimidt,
and H. E. Swanson, “Improved test of the equivalence principle for gravi-
falling cesium cold atoms in combination with an absolute tational self-energy,” Phys. Rev. Lett. 83, 3585–3588 (1999).
gravimeter, is a test of the underlying weak equivalence prin- 18
S. Schlamminger, K.-Y. Choi, T. A. Wagner, J. H. Gundlach, and E. G.
ciple. The result is the same as in the original experiment Adelberger, “Test of the equivalence principle using a rotating torsion bal-
and it is not competitive with WEP null tests. ance,” Phys. Rev. Lett. 100, 041101-1–4 (2008).
19
Finally, we have compared tests of the weak equivalence T. D. Wagner “Torsion-balance tests of the weak equivalence principle,”
principle using macroscopic proof masses with those using Class. Quantum Grav. 29, 184002-1–15 (2012).
20
A. Einstein, “Zur Elektrodynamik bewegter K€ orper,” Ann. Phys. 322 (10),
cold atoms to show that, although the experiments are com-
891–921 (1905); “On the electrodynamics of moving bodies,” in The
pletely different, there is no difference in the nature of the Principle of Relativity. Original Papers edited by A. Einstein and
tests and one should pursue the most promising ones, both in H. Minkowsky (University of Calcutta, 1920), pp. 1–34, available online
terms of sensitivity and in terms of differences in the physi- at <[Link]
cal properties between the atoms being tested. Results 2up>.
21 €
A. Einstein, “Uber das Relativit€atsprinzip und die aus demselben gezoge-
obtained with macroscopic bodies are superior in both
respects, with prospects for a breakthrough when the experi- nen Folgerungen,” Jahrbuch f€ ur Radioaktivit€at und Elektronik, 4, 411–462
(1907); “On the relativity principle and conclusions drawn from it,” in
ments will be performed in space. H. M. Schwarz, “Einstein’s comprehensive 1907 essay on relativity, part
III,” Am. J. Phys. 45(10) 899–902 (1977).
22
ACKNOWLEDGMENTS G. Holton, Thematic Origins of Scientific Thought (Harvard U.P.,
Cambridge, MA, 1973).
This work was supported by ASI (Agenzia Spaziale 23
A. Einstein, “How I created the theory of relativity,” translated by Y. A.
Italiana) and INFN, and it was performed in part at JPL, Ono, Phys. Today 35(8), 45–47 (1982).

535 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 535

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]
24 38
C. M. Will, “The confrontation between general relativity and R. H. Dicke, “Eotvos experiment and the gravitational red shift,” Am. J.
experiment,” Living Rev. Relativity 9, 1–100 (2006), available online at Phys. 28, 344–347 (1960).
39
<[Link] V. W. Hughes, H. G. Robinson, and V. Beltran-Lopez, “Upper limit for
25
R. H. Dicke, The Theoretical Significance of Experimental Relativity the anisotropy of inertial mass from nuclear resonance experiments,”
(Blackie and Son, London and Glasgow, 1964). Phys. Rev. Lett. 4, 342–344 (1960).
26 40
A. Einstein, “Die Feldgleichungen der Gravitation,” Preussische R. W. P. Drever, “A search for anisotropy of inertial mass using a free pre-
Akademie der Wissenschaften, Sitzungsberichte (part 2), 844–847 (1915); cession technique,” Philos. Mag. 6, 683–687 (1961).
41
“The field equations of gravitation,” translated by Wikisource <http:// P. Wolf, L. Blanchet, C. J. Borde, S. Reynaud, C. Salomon, and C. Cohen-
[Link]/wiki/The_Field_Equations_of_Gravitation>. Tannoudji, “Atom gravimeters and gravitational redshift,” Nature 467, E1
27
A. Einstein, “Grundlage der allgemeinen Relativit€atstheorie,” Ann. Phys. (2010).
42
(ser. 4) 49, 769–822 (1916); “The foundation of the general theory of rela- M. A. Hohensee, S. Chu, A. Peters, and H. M€ uller, “Equivalence principle
tivity,” in The Principle of Relativity (Dover, New York, 1952). and gravitational redshift,” Phys. Rev. Lett. 106, 151102-1–4 (2011).
28 €
A. Einstein, “Uber den Einfluss der Schwerkraft auf die Ausbreitung des 43
P. Wolf, L. Blanchet, C. J. Borde, S. Reynaud, C. Salomon, and C. Cohen-
Lichtes,” Ann. Phys. 35(10), 898–908 (1911); “On the influence of gravi- Tannoudji, “Does an atom interferometer test the gravitational redshift at
tation on the propagation of light,” in The Principle of Relativity (Dover, the Compton frequency?,” Class. Quantum Grav. 28, 145017-1–145017-
New York, 1952). 23 (2011).
29 44
R. C. Tolman, Relativity, Thermodynamics and Cosmology (The M. A. Hohensee, S. Chu, A. Peters, and H. M€ uller, “Comment on: ‘Does
Clarendon Press, Oxford, 1934). an atom interferometer test the gravitational redshift at the Compton
30
A. Einstein, “Ist die Tr€agheit eines K€ orpers von seinem Energieinhalt frequency?,’” Class. Quantum Grav. 29, 048001-1–5 (2012).
45
abh€angig?,” Ann. Phys. 18(13), 639–641 (1905); “Does the inertia of a P. Wolf, L. Blanchet, C. J. Borde, S. Reynaud, C. Salomon, and C. Cohen-
body depend upon its energy content?” <[Link] Tannoudji, “Reply to comment on: ‘Does an atom interferometer test the
einstein/E_mc2/www/>. gravitational redshift at the Compton frequency?,’” Class. Quantum Grav.
31
R. F. C. Vessot, M. W. Levine, E. M. Mattison, E. L. Blomberg, T. E. 29, 048002-1–3 (2012).
46
Hoffman, G. U. Nystrom, and B. F. Farrel, “Test of relativistic gravitation C. S. Unnikrishnan and G. T. Gillies, “Reexamining the roles of gravita-
with a space-borne hydrogen maser,” Phys. Rev. Lett. 45, 2081–2084 tional and inertial masses in gravimetry with atom interferometers,” Phys.
(1980). Lett. A 377, 60–63 (2012).
32 47
R. Pegna, private communication I (2012). A. M. Nobili, M. Shao, R. Pegna, G. Zavattini, S. G. Turyshev, D. M.
33
A. M. Nobili, G. L. Comandi, D. Bramanti, Suresh Doravari, D. M. Lucchesi, A. De Michele, S. Doravari, G. L. Comandi, T. R. Saravanan, F.
Lucchesi, and F. Maccarrone, “Limitations to testing the equivalence prin- Palmonari, G. Catastini, and A. Anselmi, “‘Galileo Galilei’ (GG): space
ciple with satellite laser ranging,” Gen. Relativ. Gravit. 40, 1533–1554 test of the weak equivalence principle to 1017 and laboratory demon-
(2008); Gen. Relativ. Gravit. 40, 1555 (2008). strations,” Class. Quantum Grav. 29, 184011-1–18 (2012).
34 48
L. Blanchet, C. Salomon, P. Teyssandier, and P. Wolf, “Relativistic theory R. Pegna, private communication II (2012).
for time and frequency transfer to order c3,” Astron. Astrophys. 370, 49
S. Fray, C. A. Diez, T. W. H€ansch, and M. Weitz, “Atomic interferometer
320–329 (2001). with amplitude gratings of light and its applications to atom based tests of
35
T. P. Krisher, “Parametrized post-Newtonian gravitational redshift,” Phys. the equivalence principle,” Phys. Rev. Lett. 93, 240404-1–4 (2004).
50
Rev. D 48, 4639 (1993). P. Storey and C. Cohen–Tannoudji, “The Feynman path integral approach
36
A. Schild, “Equivalence principle and red-shift measurements,” Am. J. to atomic interferometry. A tutorial,” J. Phys. II France 4, 1999–2027
Phys. 28, 778–780 (1960). (1994).
37 51
W. Rindler, “Counterexample to the Lenz-Schiff argument,” Am. J. Phys. C. S. Unnikrishnan, “The Equivalence Principle and Quantum Mechanics:
36, 540–544 (1968). A theme in harmony,” Mod. Phys. Lett. A 17, 1081–1090 (2002).

536 Am. J. Phys., Vol. 81, No. 7, July 2013 Nobili et al. 536

Downloaded 15 Sep 2013 to [Link]. Redistribution subject to AAPT license or copyright; see [Link]