PAPERS | JUNE 01 2025

Detecting gravitational waves with light

Markus Pössel

Am. J. Phys. 93, 499–510 (2025)

[Link]


View Export
Online Citation

Articles You May Be Interested In

Terrestrial very-long-baseline atom interferometry: Workshop summary

AVS Quantum Sci. (May 2024)

07 July 2025 [Link]

Detecting gravitational waves with light
€ssela)
Markus Po
Haus der Astronomie and Max Planck Institute for Astronomy, K€
onigstuhl 17, 69124 Heidelberg, Germany
(Received 16 July 2024; accepted 7 April 2025)
The strong evidence for low-frequency gravitational waves from pulsar timing arrays (PTAs),
published in 2023, has widened the scope for teaching about gravitational wave astronomy. This
article provides a simple, unified overview of the detection of gravitational waves using light
waves that encompasses the recent PTA detections, the by-now classic interferometric detections
using LIGO and similar detectors, and the yet-to-be-accomplished detections using long-arm
detectors like the space-born LISA. The presentation is at a level accessible for undergraduate
students. The influence of gravitational waves on light is derived in a way that makes use only of
basic gravitational wave properties and Einstein’s equivalence principle. # 2025 Author(s). All article
content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (https://
[Link]/licenses/by/4.0/).
[Link]

I. INTRODUCTION separated in the x and y directions (instead of changing direc-

tion over time) makes this particular example a linearly
Results published in the summer of 2023 provide strong polarized wave.
evidence for a low-frequency gravitational wave background For the following, let us concentrate on one of the paral-
from measurements using pulsar timing arrays (PTA).1–4 In lel planes, say, the plane z ¼ 0. Characteristically, test par-
January 2024, the European Space Agency gave the go- ticle distances within that plane change by a (direction-
ahead for the space-based gravitational wave detector LISA, specific) factor: If two particles are initially separated in a
slated for launch in 2037.5 Both developments provide a given direction by the distance L, a passing gravitational
challenge for teaching about gravitational waves at an under- wave will change that distance in proportion to some factor
graduate level: Neither PTA nor LISA can be understood aðtÞ, as aðtÞ  L. All separations between free-floating par-

07 July 2025 [Link]

using the so-called short-arm approximation for interfero- ticles in the same direction will vary by the same factor; an
metric detectors that is commonly employed when teaching initial distance 2L will vary as aðtÞ  2L, and so forth. For
about gravitational wave detectors.6–9 the gravitational wave shown in Fig. 1, distances in the x
The aim of this article is to provide teachers with a com- direction are changed by a factor ax ðtÞ; distances in the y
prehensive account of gravitational wave detection with direction by another factor ay ðtÞ. As one would expect,
electromagnetic radiation, that is., light, which encompasses changes in length that have both an x and a y component
not only both pulsar timing and a setup like that of LISA but can be calculated using Pythagoras’s theorem. The fact that
also detectors like LIGO, at the level of introductory physics this is a so-called quadrupole pattern can be expressed by
or astronomy courses. After a review of the basics of gravita- ax ðtÞ ¼ 1=ay ðtÞ.
tional waves in Sec. II, we deduce their influence on light The extreme weakness of gravitational waves reaching the
propagation in Sec. III. We then apply the results to different Earth makes it convenient to write
kinds of detection scenarios, demonstrating that all of them
can be understood along the same basic principles: space- 1
craft transponders in Sec. IV, pulsar timing in Sec. V, and ax ðtÞ ¼ 1 þ hðtÞ; (1)
2
interferometric detectors in Sec. VI.
with a dimensionless function hðtÞ, the gravitational wave
II. GRAVITATIONAL WAVE BASICS strain, which satisfies jhj  1. The factor 1=2 is conven-
tional, chosen for consistency with the usual linearized
Visualizations of gravitational waves typically feature the description of gravitational waves in general relativity. The
basic quadrupole pattern shown in Fig. 1, where stretching of strain encodes the relative length change,
distances in one direction on the plane always coincides with
shrinking in the orthogonal direction, and vice versa. DL 1
The pattern illustrates the effect of a passing gravitational ¼ h: (2)
L 2
wave on free-floating test particles, here arranged in a circle
with one particle in the center. The wave in question is a Since jhj  1, we will routinely discard higher than linear
plane wave: We can imagine three-dimensional space as terms in h. Notably, we can write
“sliced up” into parallel planes, which are orthogonal to the
gravitational wave’s direction of propagation (here chosen to 1 1 1
ay ðtÞ ¼ ¼ 1
 1  hðtÞ: (3)
be the z direction), with the phase of the gravitational wave ax ðtÞ 1 þ 2 hðtÞ 2
the same within each slice. Like their electromagnetic coun-
terparts, gravitational waves are transverse: test particle The specific wave in Fig. 1 is sinusoidal, hðtÞ ¼ h0 sinðxtÞ.
accelerations are orthogonal to the direction of propagation. If we want to describe the action of this particular gravita-
That the pattern of stretching and shrinking is strictly tional wave in another of the parallel planes, at different

499 Am. J. Phys. 93 (6), June 2025 [Link] C Author(s) 2025.

V 499
 In the general-relativistic formalism for describing small-
amplitude gravitational waves, this coordinate choice, with
the time coordinate defined via co-moving clocks and spatial
coordinates via distances at a given reference time, is known
as the “TT gauge.” It is one of several possible gauge choices
for describing linearized gravitational waves, and the physi-
cal effects on any of the detector configurations presented in
the following are of course independent of the chosen gauge.
For the purpose of this article, the TT gauge has the consid-
erable advantage that it allows for a derivation of gravita-
tional wave effects on light that requires little more than
Einstein’s equivalence principle: the fact that even within a
gravitational field, physics in an infinitesimal spacetime
region around an object that is in free fall is governed by the
laws of special relativity. This allows for a derivation of the
influence of gravitational waves on light that is suitable for
students who are not familiar with the basic formalism of
general relativity.

III. HOW GRAVITATIONAL WAVES INFLUENCE

LIGHT
A. Modeling Light
Fig. 1. Action of a linearly polarized, purely sinusoidal gravitational wave, Before we examine the influence of gravitational waves
propagating orthogonally to the figure plane in the z direction, on a circle of on light, let us make explicit what we mean by light in this
free-floating test particles. Shown are eight snapshots of the time evolution
of distances around a central test particle. The time sequence progresses
context. Following standard usage in astrophysics,10 we use
clock-wise. Each snapshot shows the same region of the x–y plane. “light” to refer to all varieties of electromagnetic radiation,
not just the more specific “visible light.” Electromagnetic
radiation, in turn, is a quantum phenomenon, reaching our

07 July 2025 [Link]

detectors and interacting with optical elements such as mir-
value of the z coordinate, the generalization is simple: All rors as a stream of photons. For the practicalities of gravita-
we need to do is replace the time variable t by the delayed tional wave detection, those quantum properties play an
time t  z=c, and consider ax ðt  z=cÞ instead of ax ðtÞ, and important role. In detectors like LIGO, the fact that light is
analogously for ay and for h. This construction shows clearly reflected at mirrors not as a smooth and continuous energy
that our wave is propagating at the speed of light c in the flow, but as the stochastic rat-a-tat of photons, is responsible
positive z direction. for part of the noise that makes gravitational wave signals
Before we can examine the influence of our gravitational hard to detect. This “quantum noise,” together with the ther-
wave on light propagation, we need to give some thought to mal noise associated with thermal fluctuations, defines the
suitable coordinates. Imagine that our plane is densely filled “noise floor” that fundamentally limits the sensitivity of a
with freely floating test particles, all at rest relative to each detector design.11 The latest generation of ground-based
other prior to the arrival of the gravitational wave. Each par-
detectors goes so far as to use so-called “squeezed light,”12
ticle’s world line has an associated proper time: duration as
manipulating the quantum properties of radiation in a way
measured by a co-moving ideal clock. Before the gravita-
that suppresses the associated noise in a way that cannot be
tional wave makes itself felt, we assume spacetime geometry
to be governed by special relativity, and we synchronize the described by classical physics. Such noise estimates, how-
co-moving clocks accordingly. Once the gravitational wave ever, and even more so nonclassical light, are beyond the
has arrived, we continue to assign to each event E in our scope of the present article.
plane the time t shown by the co-moving clock of the free- Furthermore, even for the classical electromagnetic field,
floating particle that is present at E. we do not require the full description in terms of electric and
Similarly, before the gravitational wave arrives, we assign magnetic field vectors. Instead, we model interference
to each of our family of free-floating particles Euclidean x, y effects in a simplified way that is commonly used when
coordinates corresponding to its position, with the central teaching about the basics of interference and interferome-
particle in our circle as the spatial origin. We keep those try:13 electromagnetic waves are modeled as scalar waves,
coordinates for each particle fixed (“comoving coordinates”) usually taken to be sinusoidal, characterized at each location
even during the passage of the gravitational wave, and to by a value for the displacement and a phase. For such a sim-
each event E in our plane, we assign the coordinates x, y of plified wave, the displacement can be interpreted as repre-
the unique free-floating particle from our family that is pre- senting the component of the electric field vector in the
sent at E. Inter-particle distances calculated via Pythagoras’s direction of polarization for a plane, linearly polarized elec-
theorem in these co-moving coordinates only correspond to tromagnetic wave. In that last respect, at least, the model is
physical distances in the absence of gravitational waves. In rather close to reality: Plane waves of this kind are particu-
the presence of gravitational waves, we need to multiply x larly suitable for interferometry, and in detectors like LIGO,
and y coordinate differences with ax ðtÞ and ay ðtÞ, respec- considerable technical effort is invested in creating electro-
tively, to obtain physical distances. magnetic waves with just the right properties: linearly

500 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 500
polarized, high-intensity, singular-frequency, fundamental- gravitational waves. Once the gravitational wave arrives, the
mode (which roughly translates to: plane wave like) laser comoving coordinates can still serve as coordinates, that is,
light. as a means of assigning a tuple x, y as a unique identifier to
In situations where our analysis does not require the wave each point in our transverse plane. However, coordinate dif-
properties of light, we will resort to an even simpler picture: ferences will no longer correspond to physical distances.
We will model light pulses as point particles traveling at the Instead, as we saw in Sec. II, the gravitational wave stretches
speed of light, using the basic picture common in special and distances in the x direction by the factor ax ðtÞ, so a comoving
general relativity where light propagation is described in coordinate interval dx will correspond to a physical distance
terms of “light-like” world lines. Note that, on the quantum ds ¼ ax ðtÞ  dx.
level, there is no fundamental difference between a light We had already associated a comoving clock with each of
pulse (essentially, a bunch of photons) and the traveling the free-floating particles, and used those clocks to define
maximum of an electromagnetic wave (again, essentially, a our time coordinate. Let us go one step further and for each
bunch of photons). In our model, light pulses and the max- of the particles, imagine a comoving observer, who can per-
ima (or any other fixed-phase points) of a sinusoidal wave form basic (local) measurements. Since the particle, and
travel at the same speed, in special relativity: at the usual hence the comoving observer, are in free fall, Einstein’s
speed of light c. equivalence principle applies: For a light signal passing by at
time t, such an observer will find that, as measured by their
B. Equivalence principle and light propagation own clock (which is the same as the local time given by our
Consider a light pulse propagating in the x direction in one time coordinate t) and their local meter stick, the signal trav-
of our transverse planes. If we were in special relativity (or els at the usual constant speed c.
classical physics), we would automatically assume that a This fact allows us to find the trajectory xðtÞ of our light
pulse that starts out in the x direction will keep propagating pulse, in terms of the time coordinate t and comoving space
in the x direction. In general relativity, where light gets coordinate x we had defined: Let dt be the infinitesimal time
deflected under the influence of gravity, this statement interval, as measured by the local comoving clock, that it
should not be taken for granted, but in the special situation takes for the light to traverse the coordinate interval dx. The
we have here, symmetries guarantee that our light pulse physical distance corresponding to dx, such as our comoving
indeed keeps propagating in the x direction: The gravita- observer will measure with their local meter stick, is
tional wave’s effects are transverse, so we know our pulse ds ¼ ax ðtÞ  dx. The equivalence principle then tells us that
will not deviate out of the plane. By construction, the gravi-
tational wave’s effects are symmetric about the x axis, so ds dx dt
¼ ax ðtÞ ¼ c ) c ¼ dx:

07 July 2025 [Link]

(4)
none of the directions in which the trajectory of our pulse dt dt ax ðtÞ
could deviate within the x–y plane is preferred relative to the
others. So even while we are not in the gravity-free realm of We can readily integrate this to obtain
special relativity, the setup for our simple gravitational wave
ensures that a light pulse that starts out in the x direction will ðt
dt0
continue to propagate in the x direction (and an analogous x  xi ¼ c ; (5)
ax ðt0 Þ
statement holds for a light pulse propagating in the y ti
direction).
Wherever the light pulse passes, there will be one of the where ti is the time the light signal leaves its initial location
family of free-floating particles that we introduced in Sec. II. at comoving coordinate value xi , and t the time it arrives at
At this point, we make use of Einstein’s equivalence princi- the comoving coordinate value x. For light propagating in
ple, which encapsulates a realization Einstein had at the very the y direction, the same reasoning applies, but with ay
beginning of his path to his theory of general relativity, and instead of ax .
that he later called the happiest thought of his life: For an
observer in free fall in a gravitational field, the immediate
effects of gravity—such as the pull felt by an observer stand- C. Doppler formula
ing on the Earth’s surface—are absent. Since locally, all Next, consider the following setup: Within our family of
objects fall at the same rate, such an observer would see free-floating particles, we select two particles a and b, both
other objects floating alongside themselves. Enclose the located on the x axis, at locations xa and xb , respectively.
observer in a small cabin, and they would not be able to dis- Then, we send two light pulses in quick succession from a to
tinguish whether they were in free fall in a gravitational field,
b. Designate the time at which the first light pulse is emitted
or else in deep space, far from all sources of gravity.
by the particle a as ta , and the time of emission of the second
Einstein generalized this statement to encompass all of phys-
light pulse as ta þ dta . Conversely, denote the time the first
ics, and the result is known as Einstein’s equivalence princi-
pulse arrives at the particle b as tb , and the arrival time of the
ple: for an observer in free fall in a gravitational field, the
local physical laws are those of special relativity—at least in second pulse as tb þ dtb . From Eq. (5), we know that
an infinitesimally small neighborhood of space-time. ðtb ð b
tb þdt
For our space-filling family of free-floating particles, this dt0 dt0
means we can proceed as follows. We had introduced xb  xa ¼ c ¼c ; (6)
ax ðt0 Þ ax ðt0 Þ
comoving coordinates x and y, defined via physical distances ta ta þdta
in the absence of any gravitational wave. For a free-floating
particle on our x axis, its (permanently assigned) x coordi- since both light pulses start out at the x coordinate value xa
nate is its distance from the origin in the absence of and arrive at xb . It follows that

501 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 501
 ð b
tb þdt ðtb of emission of the light and on its state at the time of the
dt0 dt0 light’s arrival.
0
 ¼ 0: (7)
ax ðt Þ ax ðt0 Þ We can also express the change in terms of the period P of
ta þdta ta
our simple sinusoidal light wave. P is the time interval between
the arrival of two consecutive maxima, and thus is P ¼ k=c.
The limits of the first integral can be rewritten as
The relative change of that period due to a gravitational wave is
ð b
tb þdt ðta ðtb ð b
tb þdt ðtb ð a
ta þdt ð b
tb þdt
DP 1
¼ þ þ ¼  þ : (8) ¼ z ¼ ½hðtb Þ  hðta Þ: (15)
P 2
ta þdta ta þdta ta tb ta ta tb
The same formula applies to any periodic signal, and we will
In all the situations we will consider in the following, dta and revisit it when we consider pulsar timing in Sec. V. For the
dtb will be short compared with the time scale for any change same sinusoidal light wave, its frequency f ¼ 1=P and wave-
of ax ðtÞ, which means we can use the mean value theorem length k are linked by f  k ¼ c, so
for definite integrals, Df Dk 1
¼ ¼ z ¼ ½hðta Þ  hðtb Þ: (16)
ð a
ta þdt f k 2
dt0 dta
0
 ; (9) This version will become important as we consider the
ax ðt Þ ax ðta Þ
ta (as yet unrealized) detection of gravitational waves using
space probes in Sec. IV.
and similarly for the tb integral. Putting everything together,
(7) is transformed to D. Phase formula
dta dtb In addition to the various kinds of Doppler effect described
¼ : (10)
ax ðta Þ ax ðtb Þ in Sec. III C, we can use the light propagation Eq. (5) to deduce
phase information for sinusoidal light waves. Later on, we will
Evidently, the distances between successive light pulses want to describe interferometers such as LIGO, we will model
change in the same way as the distances between our free- elements such as light sources, beam splitters, mirrors, and
floating particles: in proportion to ax ðtÞ. detectors, as free-floating particles, whose distances from each
Now, instead of two successive light pulses, consider a other change under the influence of a passing gravitational

07 July 2025 [Link]

sinusoidal light wave propagating from particle a to particle wave. We will restrict our analysis to the simplest case, when
b. We can choose dta to be the time interval between the all these elements are in the same plane, transverse to the direc-
emission of one wave crest and the next, which means that tion of propagation of the gravitational wave, and where light
dtb is the time interval between the arrivals of those two propagates as a sinus wave either in the x or in the y direction.
wave crests at particle b. In terms of the wavelength ka as For simple sine waves like our light waves, time t and
the wave is emitted at a and the wavelength kb as it arrives at phase u are related as u ¼ 2pft þ u0 ; with f the frequency of
b, we then have ka ¼ c  dta and kb ¼ c  dtb , respectively, the wave. Thus, to relate phases at different locations—say,
and (10) becomes the phase at the light source and at a distant detector—we
will need to be able to tell how much time our light requires
ka kb to travel from a starting point to an end point.
¼ : (11)
ax ðta Þ ax ðtb Þ For travel along the x direction, this amounts to asking: At
what time t will a light signal that has left the starting point
In terms of the gravitational strain h, at x0 at a time t0 reach the coordinate value x? As it stands,
Eq. (5) provides an answer to the converse question, namely
kb 1 þ 12 hðtb Þ 1 the coordinate value x the light has reached at the time t,
¼  1 þ ½hðtb Þ  hðta Þ: (12)
ka 1 þ 12 hðta Þ 2 when previously at time ti it was at xi ,
ðt ðt
The relative wavelength shift between emission and arrival dt0 dt0
is usually called z, which in astronomy goes by the name x ¼ xi 6c ¼ x i 6c ; (17)
aðt0 Þ 1 þ 12 hðt0 Þ
“redshift” (blueshifts are called “negative redshifts” in this ti ti
context), and is
where the signs correspond to propagation in the positive
kb  ka and negative x direction, respectively. Our next task will be
z : (13)
ka to find an approximate solution for t in terms of x.
Discarding terms higher than linear order in h, (17) can be
From Eq. (12), we have rewritten as
Dk 1 ðt
 z ¼ ½hðtb Þ  hðta Þ: (14) c
k 2 x ¼ xi 6cðt  ti Þ7 hðt0 Þ dt0 : (18)
2
ti
This is a key result for the “Doppler shift” induced by a grav-
itational wave: z depends both on the gravitational wave’s For a better understanding of this equation, consider the rate
state (i.e., its amplitude, orientation, and phase) at the time of change,

502 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 502
  
dx 1 Hubble-flow galaxies will measure the same value for the
¼ 6c 1  hðtÞ : (19) mean density of the universe around them.14
dt 2
Comoving coordinates in an expanding universe can be
This is the “coordinate speed of light,” that is, the rate at introduced using that same family of idealized Hubble-flow
which the x location of a light pulse or light wave, as galaxies. Consider a snapshot of the universe at some fixed
expressed in our comoving coordinates, changes with t. In cosmic time t0 . Spatial relations within that snapshot can be
the absence of a gravitational wave (i.e., for hðtÞ ¼ 0), both described using suitable coordinates; in the simplest case,
our x coordinate and our t coordinate revert to the usual coor- that of a spatially flat universe, we could assign to each gal-
dinates of an inertial system in special relativity, so light axy standard Cartesian coordinates. The only change in spa-
moves at the speed c. The presence of the hðtÞ term introdu- tial relations relative to that t ¼ t0 snapshot will be one of
ces a small variation of the coordinate speed over time. Note overall scale. That allows us to assign each galaxy its snap-
that, since jhj  1, the overall sign of this expression does shot coordinate at t ¼ t0 as a permanent, “comoving” coordi-
not change: Light moving either in the positive or negative x nate, valid at arbitrary times t. The only drawback is that for
direction does not change direction as the gravitational wave t 6¼ t0 , the distances associated with those spatial coordinates
passes. do not correspond to physical distances. That is easily
As a first step toward solving (18) for t, re-write (18) as remedied, though: We can go from comoving distances
(which, after all, correspond to physical distances at t ¼ t0 )
ðt to physical distances at any time t, simply by rescaling with
x  xi 1
t ¼ ti 6 þ hðt0 Þ dt0 : (20) aðtÞ=aðt0 Þ.15
c 2
ti The parallels with the description of gravitational waves
in Sec. II should be evident, and the simplified calculation
From Eq. (19), we know that the function xðtÞ is strictly for light propagation based on the equivalence principle
monotonic, hence invertible. Retaining only terms that are works in either case: Our family of free-floating test particles
first-order in hðtÞ in expression (20), we replace the integra- correspond to Hubble-flow galaxies; the time coordinate we
tion limit by t  ti 6ðx  xi Þ=c, and make a change of varia- defined corresponds to cosmic time, and both situations can
bles from t0 to x0 using dt0 ¼ 6dx0 =c, to obtain be described using comoving coordinates. The direction-
xxi
dependent scale factors ax ðtÞ and ay ðtÞ correspond to a sin-
ti 6
ðc gle, universal scale factor aðtÞ in cosmology, which governs
x  xi 1
t ¼ ti 6 þ hðt0 Þ dt0 (21) cosmic expansion. The Doppler formula (11) yields the
c 2 standard cosmological redshift, and the analog of Eq. (5)

07 July 2025 [Link]

ti
describes light propagation in an expanding cosmos. In the
ðx   end, both seemingly different situations are governed by the
x  xi 1 x 0  xi
¼ ti 6 6 h ti 6 dx0 : (22) same physics, rooted in general relativity.
c 2c c
xi

IV. GRAVITATIONAL WAVE DETECTION WITH

Let us call this the phase formula for light propagation influ-
TRANSPONDERS AND SPACE PROBES
enced by a gravitational waves. It is the counterpart to the
Doppler formula in its various guises (14)–(16). We can get The basic setup for our first example is shown in Fig. 2: an
from one to the other by integrating or differentiating. antenna on Earth (left) sends a radio signal with frequency f
Which formula best describes how gravitational waves are to a distant space probe (right); the probe’s transponder
detected will depend on the setup. Doppler and phase for- immediately sends the radio signal back. We call the Earth-
mula are two sides of the same physical coin. to-probe distance D ¼ cT=2; with T the total two-way travel
time.
E. Relation to cosmology Continuing with the simplified setup we concentrated on
From a pedagogical point of view, the parallels between in Sec. II and Sec. III, we only consider plane gravitational
light propagation under the influence of a gravitational waves propagating orthogonal to the radio signal. For the
wave and in an expanding universe are worthy of note. Doppler shift due to the gravitational wave, it doesn’t matter
Recall that, in the standard description of an expanding whether the radio signal is traveling in the positive or in the
universe, distances between distant galaxies are propor- negative x direction. The total effect will only depend on the
tional to a time-dependent cosmic scale factor aðtÞ. The time the signal was emitted and on its total travel time; while
pattern of motion this imposes on galaxies is called the it is traveling, the signal will be Doppler-shifted regardless
Hubble flow. We can introduce a space-filling family of of whether it is moving to the left or to the right. Back on
point-like, idealized galaxies, “fundamental particles of Earth, we monitor the frequency of the returning radio sig-
cosmology.” whose motion follows the Hubble flow nal. By the Doppler formula (16), the relative frequency shift
exactly. The usual cosmological time coordinate can then at reception time t is
be defined with reference to that family: At each point in
space, cosmic time is time as measured on the comoving
clock of the local Hubble-flow galaxy. Global synchroni-
zation for those comoving clocks is provided by making
use of the homogeneity of the universe in these cosmolog-
ical models: the comoving clocks are synchronized so that Fig. 2. Basic setup for detecting a gravitational wave using a spacecraft with
at any given moment in cosmic time, all observers in transponder.

503 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 503
 Df 1
¼ ½hðt  TÞ  hðtÞ: (23)
f 2

Measuring the difference hðt  TÞ  hðtÞ instead of hðtÞ

directly requires extra analysis, since there are two contribu-
tions to the Doppler shift, from hðtÞ and hðt  TÞ. However,
there is an interesting type of signal where the analysis is
straightforward: a transient gravitational wave signal, like
the chirp signal from two merging black holes. Assume that
this transient gravitational wave begins to make its influence
felt in the transverse plane where Earth, the space probe, and
the radio signal are located, at time ta , and that the gravita-
tional-wave-induced changes are complete by the time tb , Fig. 3. Sensitivity of a transponder setup for a probe near Neptune,
T ¼ 30 000 s, as a function of gravitational wave frequency.
more specifically: that hðtÞ ¼ const: (although not necessar-
ily zero!16) for t > tb .
In the Doppler shift formula (23), there are two “copies” period is such that t and t  T correspond to identical phases
of that signal: first hðtÞ and later on, delayed by a time inter- of the wave, the difference is identically zero. To gravita-
val T, the term hðt  TÞ. If the delay T is large enough, tional waves with frequency fgw ¼ n=T; n 2 N, our setup is
tb  ta < T, then the hðtÞ copy of the signal will have run its completely insensitive!
course before the term hðt  TÞ has even begun to diverge Figure 3 shows a logarithmic plot of the dependence of
from zero. In that case, we see two copies of the transient ðhc Þ2 on fgw . The T was chosen so as to roughly correspond
gravitational wave signal cleanly separated from each
to a probe near Neptune, at about 30 au, with a round-trip
other: first the transient signal in hðtÞ, then a pause where
time of roughly 8 hours. This is our first encounter with a
nothing happens, and then a repeat of the signal via the term
general feature of gravitational wave detectors: linear growth
hðt  TÞ.
The basic principles for this kind of detection were due to sinðxÞ  x in Eq. (24) as long as the detector time
worked out half a century ago.17,18 Since then, transponder scale is small compared to the gravitational wave period, fol-
measurements in search of gravitational waves have been lowed by periodic sensitivity drops where the gravitational
made using a number of interplanetary spacecraft, notably wave period is an integer multiple of the detector time scale.
the Cassini probe. So far, those measurements have not Sensitivity curves in the literature, e.g., Fig. 2 in Ref. 20,
will look similar, but usually not identical—they average

07 July 2025 [Link]

been sensitive enough to detect gravitational waves.
However, whenever gravitational waves are searched for at over different orientations of the detector relative to the
a given sensitivity, but not detected, the results provide an gravitational wave, whereas we only consider the orthogonal
upper limit for the strength of gravitational wave signals in case.
the frequency range that was covered by the measure-
ments.19 A positive result, that is, a direct detection of V. PULSAR TIMING ARRAYS
gravitational waves using this method, is likely to take
Next, consider the case where our radio signal is not arti-
another decade. Planned space missions to the ice giants,
ficial, but the highly regular series of pulses reaching us
with a travel time of about 10 years, would provide an
from a distant pulsar. The situation is sketched in Fig. 4.21
opportunity for these kinds of measurement.20 Detections
The pulses we receive on Earth are created through a light-
should be feasible for gravitational wave frequencies
house effect: The pulsar emits intense radio waves in two
between about 105 and 1 Hz. Between the start of such a
opposite directions. Those directions are not aligned with
mission around 2030 and its arrival at Uranus or Neptune
the pulsar’s rotational axis, and thus trace out two cones as
about 10 years later, transponder measurements might
detect merging supermassive black holes, a stellar-size the pulsar rotates. If the pulsar is aligned just right, we
black hole merging with a supermassive black hole receive a radio pulse every time the pulsar’s beam brushes
(extreme mass ratio inspiral, EMRI), or merging stellar- across Earth.
mass black holes. For T the (one-way) travel time of the signal from the pul-
Substituting hðtÞ ¼ h0  sinð2pfgw tÞ; we can estimate the sar to Earth, the distance is D ¼ cT. Pulse times-of-arrival
sensitivity of such a transponder-detector to monochromatic (TOA) are recorded by radio telescopes utilizing precise
gravitational waves with various frequencies fgw . By the atomic clocks. Those TOA measurements require summing
addition theorem for two sine functions, the relative fre- up a considerable number of consecutive pulses in a coherent
quency shift (23) at reception time t is way, and fitting a suitable template for the shape of the pulse
to the result. Uncertainties in this fitting procedure are the
Df h0   main source of uncertainty for TOA measurements. The
¼ sinð2pfgw ½t  T Þ  sinð2pfgw tÞ TOA sequence is then decomposed into a regular part,
f 2
 
¼ h0 cos 2pfgw ½t  T=2 sinðpfgw TÞ : (24)
|fflfflfflfflfflfflffl{zfflfflfflfflfflfflffl}
hc

This is a sine wave with an amplitude modulation h0  hc .

When hc is zero, our transponder is not able to detect the
gravitational wave at all. This is because what we measure is Fig. 4. Basic setup for the detection of a regular pulsar signal. The direc-
the difference hðt  TÞ  hðtÞ: If the gravitational wave tional radio signals from the pulsars are shown as cones.

504 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 504
corresponding to the period P, the correction due to (con- ðt
_ and a time-dependent correction DP h0
stant) period drift P, r sin ð2pf  t0 Þdt0 (27)
that changes the time interval between the arrival of each 2
t0
pulse and its successor. In the following simplified toy
model, we will instead pretend that our radio telescope is h0
determining times-of-arrival for single, separate pulses, that ¼ ½cosð2pf  t0 Þ  cosð2pf  tÞ; (28)
4pf
we have already determined P in the absence of gravitational
waves, and that P_ ¼ 0, so that we can now set out directly to demonstrating how the detection is more sensitive for
measure the influence of the gravitational wave by consider- smaller gravitational wave frequencies f. Figure 5 shows
ing the DP. how this timing residual amplifies the effect of the individual
The most stable known pulsars are millisecond pulsars, period shifts up to a maximum that consists of all the posi-
with rotation periods P on the scale of milliseconds. Those tive shifts DP of the sine wave adding up.
pulsars have been spun up to their high rotational speed by This leaves us with one remaining problem: We neglected
accreting material from a companion star, and P is three to the delayed term hðt  TÞ in Eq. (25). We can solve the prob-
four orders of magnitude more stable than for the much lem by considering not a single pulsar, but a larger set of
more common younger pulsars with rotation periods of pulsars: a pulsar timing array (PTA). These pulsars will nat-
about a second.22 From the Doppler formula (15), it follows urally be at different locations in space. In our one-
that the period shift for a pulsar signal arriving at time t that dimensional toy model, the pulsars would be at different dis-
has traveled to us along the x axis is tances in the x direction, with different pulse travel times Ti .
Adding up their various timing residuals, we obtain
DP 1
¼ ½hðtÞ  hðt  TÞ: (25) ð t
P 2 X
N
1  
rPTA ¼ hðt0 Þ  hðt0  Tj Þ dt0
What we can measure is once more the difference of two j¼1
2
t0
terms: the state of the gravitational wave hðtÞ as the pulse
arrives at Earth and its state hðt  TÞ at the time the pulse ðt XN ðt
N 0 0 1
left the pulsar. ¼ hðt Þdt  hðt0  Tj Þdt0 : (29)
For the moment, let us ignore hðt  TÞ, focus on hðtÞ, and 2 j¼1
2
t0 t0
consider the accuracy needed for detecting DP. The current
PTA measurements detect a “stochastic background,” the The first term of the sum, the “Earth term” hðtÞ describing

07 July 2025 [Link]

combination of many unresolved gravitational wave events the phase of the gravitational wave at the time the pulses
in various locations in the cosmos. In our toy model, we simultaneously arrive on Earth, has received a boost in our
replace this background by a continuous sine wave with simplified situation: it is amplified by the number N of pul-
maximum amplitude h=2  1014 . For a millisecond pulsar, sars in the set. In the remainder, each part of the sum is a
say P ¼ 5 ms, this would mean DP ¼ 5 108 ns. Even in a “pulsar term,” describing the state of the gravitational wave
fictitious world where we could determine pulse times-of- at the event when the pulse left a particular pulsar. The
arrival with an accuracy of 1 ns, this order of magnitude pulsar term contributions to the sum will have different
would be impossible to detect. phases—some will be positive, some negative. Those terms
However, there is hope. Let us, just for the moment, average out, so the transition from a single pulsar to an array
consider DP > 0 as constant over the observation time— of pulsars has indeed solved our problem.
corresponding to an extremely low-frequency gravita- For a realistic three-dimensional pulsar timing array, the
tional wave. Using our clock to document pulse arrival analysis is more complicated, but an important part of what
time, we find the second pulse is DP later than expected. is going on is covered by the toy model: There, too, docu-
The third pulse will be late by 2  DP, and the ðn þ 1Þth menting timing residuals over a sufficiently long time will
pulse by n  DP. Even where DP itself is undetectable, a boost the signal. The more complicated geometry, with
shift n  DP with large n need not be. In our example, after the pulsars distributed all over the sky, means that the toy
500 days (about 4 107 seconds, or n ¼ 8 109 periods),
the pulse is late by 400 ns, a shift that current TOA mea-
surement techniques, whose accuracy can be better than
100 ns, would be able to detect.
Returning to non-constant DP, we define the timing resid-
ual r as the cumulative shift at time t. For P sufficiently
small over our measuring interval from t0 to t, we can write
the N summed-up individual period shifts as an integral,

X
N X
N ðt
1 1
r¼ DPðti Þ ¼ hðti Þ  P  hðt0 Þdt0 : (26)
i¼1 i¼1
2 2
t0

To see how this works, consider the simple example of a

monochromatic (sine) gravitational wave with frequency fgw ,
amplitude h0 ¼ 0:1 and 1=fgw ¼ 40 P. For such a signal with Fig. 5. Timing residuals for a sine gravitational wave with 1=fgw ¼ 40 P and
hðtÞ ¼ h0 sinð2pf  tÞ, the unrealistically large amplitude h0 ¼ 0:1.

505 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 505
model’s simple adding up of residuals (29) is not sufficient. let us describe the amplitude of the light wave (e.g., the electric
We require an additional step: First, pairs of pulsar signals field in z direction) emanating from BS into each arm by a sine
are correlated. Then, those correlations are averaged, which function A  sinð2pf  tÞ, with f the frequency of the laser light.
has the same effect as in Eq. (29) of the pulsar terms averag- For the remainder of this section, assume a constant length
ing out. The usual averaging-out procedure requires the fact offset, with BSM1 ¼ L and BSM2 ¼ L þ DL. Ignoring
that the real gravitational wave background is not a sine discrete phase jumps associated with reflection, the light
wave from a well-defined direction, as in our toy model, but arriving back at BS, which makes its way to the detector D,
a stochastic mix of signals reaching us from all possible is then given by
directions in the sky.23,24 This kind of analysis is the basis
for the June 2023 announcements of various PTA collabora- A
sinð 2pf ½t  2L=cÞ þ sinð 2pf ½t  2ðL þ DLÞ=cÞ ; (30)
tions.1–4 If instead of a stochastic background signal, the 2
array were to look at a monochromatic signal from a local-
ized source, then at least for nearby sources, a more compli- a coherent superposition that takes into account the different
cated PTA analysis could make use of both the Earth term round-trip times for light in the two arms. With the sum for-
and the pulsar term to reconstruct hðtÞ.25 For a transient mula for sine functions, this is
instead of a monochromatic signal, the same reasoning
would apply as for the transponder method in Sec. IV: given A sinð 2pf ½t  2L=c  DL=cÞ  cosð2pDL=kÞ; (31)
a sufficiently brief transient signal, we would first detect the
Earth term, and then after a pause (which, depending on the where we have used f  k ¼ c for the light’s frequency f and
pulsar distance, would likely defy the time scale of any wavelength k. The first factor in this product is the same as
human research project), the pulsar term. the original form of our light wave, except for a constant
phase shift. A photodetector integrates light power over
numerous periods of oscillation, so this term does not influ-
VI. LIGO, LISA ET AL
ence our measurement at all. The second factor modulates
A. Michelson interferometer basics the light signal as a whole, and its square will determine how
much power we observe. If DL ¼ 0 were the default state
The basic setup of an interferometric detector like LIGO and DL 6¼ 0 what we are trying to detect, a cosine, or the
or Virgo is that of a Michelson interferometer as in Fig. 6. square of a cosine once we look at the brightness of the light,
Monochromatic light from a light source S propagates to a has two disadvantages. One is that DL ¼ 0 corresponds to a
beam splitter BS; half of the light takes a round trip via the maximum, so detecting a gravitational wave would amount
mirror M1, the other via the mirror M2. On returning to BS,

07 July 2025 [Link]

to detecting a small brightness change of already rather
the light portions are combined coherently, and some portion bright light. It is easier to detect a small change in the dark,
of combined light reaches the photodetector D. In a detector and thus advantageous to adjust the detector so that DL ¼ 0
like LIGO, the optical elements are suspended so that we can corresponds to complete dark or near-dark.26 Shifting Mirror
treat their motion, at least when it comes to the distances 2 by k=4 transforms the cosine into a sine,
between BS and M1/M2, as free-fall motion. That is fortu-
nate: For our simple linearly polarized gravitational wave cosð2p½ DL  k=4=kÞ ¼ sinð2pDL=kÞ: (32)
propagating orthogonal to the detector plane, we can use the
formalism we developed in Secs. II and III. Done? No, since the new sine-square function for the bright-
Let BS and M2 be separated in the x direction, BS and M1 ness still has an extremum at DL ¼ 0, which means its
in the y direction. When the two portions of light leave the response to DL 6¼ 0 is a second-order effect! An additional
beam splitter BS, they have the same phase. For concreteness, small length change in one arm, corresponding to a phase
shift D/, helps. With the shifted argument
DL
v  2p þ D/; (33)
k
we have
DL
sin2 v  sin2 ðD/Þ þ sinðD/ÞcosðD/Þ4p : (34)
k
Now the change in the power signal we measure is directly
proportional to the length change DL. We will take this as
the operating principle of our interferometric detector.27 Our
take-away is: Whenever we end up with our light signal
modulated by a cosine as in Eq. (31), suitable constant phase
shifts can make it so that the argument of that cosine is pro-
portional to the gravitational wave signal we detect.

B. Interferometric GW detectors
Fig. 6. Basic setup for a Michelson interferometer influenced by a gravita- Now, we extend our analysis to length changes caused by
tional wave. gravitational waves. We again consider the layout shown in

506 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 506
Fig. 6, but this time with gravitational-wave-induced length about the travel back and forth; for calculating the travel
changes given by Eqs. (1) and (3), respectively. time, the result for traveling from 0 to L and back is the same
Just as in Sec. VI A, in order to determine the action of the as for traveling in a single direction from 0 to 2L. All in all,
gravitational wave on our interferometer, we need to deter- for light traveling from BS to M2 and back along the x direc-
mine when light that is returning to the beamsplitter now, at tion, we have
time t, originally started out at the beamsplitter—which will
tell us the relative phase of the two portions of light arriving
2ðL  
2L 1 x0  2L
at the beamsplitter simultaneously, at this time (and, in part, ti ¼ t   h tþ dx0 : (35)
going on to the photodetector). With the help of Eq. (22), we c 2c c
0
can express the beamsplitter start time ti in terms of the
arrival time. An advantage of the simplicity of our situation, The corresponding formula for the y arm is obtained via
where light travels orthogonally to the propagation of the h7!  h. Coherent superposition amounts to adding up the
gravitational wave, is that we do not need to be worried two sine contributions from the two arms, as

0 2 31 0 2 31
ð
2L  0
 ð
2L  0

B 6 2L 1 x  2L 7C B 6 2L 1 x  2L 7C
sin@ 2pf 4t   h tþ dx0 5A þ sin@ 2pf 4t  þ h tþ dx0 5A
c 2c c c 2c c
0 0
0 1
  2ðL  
2L Bpf x0  2L C
¼ 2 sin 2pf t  cos@ h tþ dx0 A: (36)
c c c
0

This is the same kind of cosine term as in Eq. (31), and using side, we have substituted the time-dependent length change
the constant length offsets described in Sec. VI A, our detec- in the interferometer arm L caused by the gravitational sine

07 July 2025 [Link]

tor response will be proportional to the argument of the wave. Comparison with the argument of the cosine in Eq.
cosine. For a sine-shaped gravitational wave with hðtÞ ¼ (31) shows that this is indeed the expected result for a
h0  sinð2pfgw tÞ; that argument is Michelson interferometer where the difference between arm
lengths is 2DLðtÞ, with the time-dependent length change
2ðL  
pf x0  2L governed by hðtÞ as in our basic formula (2) for the physical
h tþ dx0 ; (37) meaning of the strain. This is known as the short-arm
c c
0 approximation, since re-written in terms of the gravitational
wave’s wavelength kgw ¼ c=fgw and the arm length L, the
which we can rewrite as Taylor approximation is valid for L  kgw :
Beyond that approximation, the amplitude of our detector
ðt response is governed by the factor
pf hðt0 Þdt0
1
t2L=c sinðpfgw sL Þ: (40)
fgw
ðt
¼ pfh0 sin ð2pfgw t0 Þdt0 In Eq. (38), let us divide this by sL to make the expression
t2L=c dimensionless; this amounts to dividing out a linear overall
  “longer arm length is better” factor. Discarding the overall
h0 f 2L sign, consider the dimensionless function
¼ cos 2pfgw t   cosð2pfgw tÞ
2 fgw c
f   1
¼ h0 sin 2pfgw ½t  sL =2 sinðpfgw sL Þ: (38) mðfgw Þ ¼ sinðpfgw sL Þ ; (41)
fgw fgw sL

In the last step, we have introduced the detector time scale which is plotted in Fig. 7 as a function of fgw sL . Evidently,
sL  2L=c as a measure of light travel time inside the detec- as long as 1=fgw is considerably larger than the light travel
tor. An interesting limiting case is fgw sL  1: Taylor- time s, the function mðfgw Þ is almost constant. This is
expanding the result (38) in fgw sL , we have where the short-arm approximation holds. Once the two
become comparable, we have periodic changes between
2p DLðtÞ maximum sensitivity and sensitivity zero, overlaid by a
L h0 sinð2pfgw tÞ ¼ 4p ; (39) downward trend—a “long-arm penalty” where, at odds
k k
with the usual “bigger is better” philosophy for such
where we have substituted the wavelength k ¼ c=f of the detectors, our detector becomes less sensitive because of
light and, on the right-hand side, where on the right-hand its longer arms.

507 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 507
 ruler to detect gravitational waves?”29,30 Light waves being
doppler-shifted is what all the detection methods have in
common. It is the Michelson interferometer that is the odd
one out, in that changes in overall path length play an impor-
tant role. The short-arm approximation, neglecting the
Doppler shifts altogether, is an extreme limiting case of a
more general setup.
Almost all of the technical efforts in constructing detectors
like LIGO, Virgo, KAGRA or LISA are directed toward the
suppression of various sources of noise that would otherwise
drown out the exceedingly weak gravitational wave signals.
It is worth noting that some of the resulting design modifica-
tions do affect the applicability of the simple interferometer
Fig. 7. Sensitivity of an interferometric detector as per (41).
model from Sec. VI B: Consider the key dimensions of the
German-British detector GEO600,31 the LIGO detectors,32
So what is happening here? There are two overall effects. and LISA33 summarized in Table I. For the ground-based
One is that light travel time in each arm is changed as detectors, GEO600 with its folded light path bringing the
the overall arm length changes due to the influence of the travel length to and from the mirrors to twice the 600 meter
gravitational wave. The other is that the wavelength of the overall length that is part of the detector’s name, ensures that
traveling light is affected directly while the light spends the short-arm approximation is safely applicable.
propagating in the detector. The short-arm approximation is The LIGO detectors, however, are not the simple
valid when the time light spends inside the detector is short Michelson interferometers of Fig. 6. LIGO boosts sensitiv-
relative to the gravitational wave period. In this case, the ity by building each arm as a Fabry–Perot interferometer:
wavelength change given by the Doppler formula (14) is with a probability of nearly 99%, light heading back in the
small, and can be neglected; the only remaining effect is that direction of the beam splitter will be reflected at an extra
of the arm length difference at that particular time. In that mirror at the inner end of the arm, taking another turn up
scenario, the detector constantly uses new, “fresh,” unaf- and down that arm. Most light will spend considerably lon-
fected light to map its differential arm length change—and ger than 27 ls in an arm. This increases laser power by a
as one would expect, the result (39) is the same as if we were factor Garm ¼ 270, corresponding to an average light stor-
dealing with a classical Michelson interferometer, its light age time s ¼ 3:6 ms, the same as for a detector with an arm
completely unaffected, but its arm lengths changing in the

07 July 2025 [Link]

length of Leff ¼ 270:4 ¼ 1080 km.
characteristic quadrupole pattern. This would seem to be well within LIGO’s sgw range, sug-
Beyond the short-arm approximation, both travel time and gesting that the short-arm approximation is not applicable.
the Doppler effects need to be accounted for, as in Eq. (40). However, it’s not that simple, either: Different portions of
The periodic points of complete insensitivity in that formula, laser light will spend a different amount of time in a LIGO
and Fig. 7, can be made plausible, as well: When s  fgw ¼ 1, arm, largely averaging out the L-specific effects in Eq. (38).
light propagating once throughout the detector experiences Due to this averaging out, the short-arm approximation
each phase of the gravitational wave exactly once, in a works well for LIGO (and Virgo, and KAGRA) after all, and
completely symmetric fashion: for every stretching of the is commonly used as the basis for analyzing the performance
distance, the corresponding shrinking; for every Doppler of such detectors.34 Discussions of the limits of the short-
stretching of the waves themselves, a corresponding blue- arm approximation and the need for corrections have a long
shift. A simplified animation of what happens outside the history within the community.35–41
short-arm regime can be found in Ref. 28. For LISA, there is a different complication. With its three
satellites forming a gigantic, approximately equilateral trian-
VII. DISCUSSION gle 2.5 106 km a side, LISA is clearly beyond the short-
For pedagogical reasons, throughout our analysis in arm approximation, and the “arm length penalty” imposed
Secs. II–VI, we have concentrated on certain special cases. by Eq. (38), with a periodic structure overlaid with overall-
A comprehensive description would need to include all possible worsening sensitivity at higher frequencies, can be clearly
gravitational wave polarizations, more general wave forms, seen in sensitivity plots.33
non-planar gravitational waves, and arbitrary orientations of the However, in practice, LISA’s performance is not directly
detectors and of the propagating light relative to the gravita- based on the phase formula (22). Instead, LISA’s laser sig-
tional wave. However, even the simplified accounts of the nals are compared to a reference laser whenever they arrive
various detection methods show the connections between seem- at one of the spacecraft (by superimposing both signals and
ingly different measurements such as pulsar timing residuals measuring the beat frequency), yielding Doppler shifts (16).
and interferometric phase comparisons. At the most fundamen-
tal level, all of the measurements that rely on gravitational Table I. Detector dimensions and frequency ranges.
waves’ interaction with light and (approximately) free-fall par-
ticles can be understood in terms of the Doppler formula in its sgw ¼ 1=fgw fgw
Detector L s ¼ 2L=c range range
various guises (14)–(16) and the phase formula (22), the former
the differential forms of the latter. GEO 600 2  0:6 ¼ 1:2 km 8 ls 0:2…20 ms 50 Hz…6 kHz
The unified view presented here is almost at the opposite LIGO 4 km 27 ls 0:5…50 ms 20 Hz…2 kHz
end of the spectrum from a question like “If light waves are LISA 2:5 106 km 17 s 10…105 s 0:1 mHz…0:1 Hz
stretched by gravitational waves, how can we use light as a

508 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 508
 14
In the analysis, the Doppler shifts are integrated up, in effect Wolfgang Rindler, Sec. 16.2 in Relativity: Special, General and
making the transition from the Doppler formula (16) back to Cosmological (Oxford U. P., New York, 2001).
15
Sec. 16.4 in Rindler, op. cit.
the phase formula (22), but with a twist: For the integration, 16
That, for transient signals such as black hole chirps, hðtÞ does not return to
artificial time delays are introduced, yielding the phase for- zero afterward, but to another (very small) constant value is known as
mula not for LISA (whose arm lengths vary by about 1% “gravitational wave memory.” Cf. Kip S. Thorne, “Gravitational-wave
over the course of a year) but for a virtual Michelson inter- bursts with memory: The Christodoulou effect,” Phys. Rev. D 45(2),
ferometer with equal-length arms. This process is called 520–524 (1992). and references therein.
17
time-delay interferometry (TDI), and it is again related to William J. Kaufmann, “Redshift fluctuations arising from gravitational
waves,” Nature 227(5254), 157–158 (1970).
noise suppression: In an equal-arm-length interferometer, 18
Frank B. Estabrook and Hugo D. Wahlquist, “Response of Doppler space-
noise due to the unavoidable jitter in the laser frequency can- craft tracking to gravitational radiation,” Gen. Relativ. Gravitation 6,
cels out. Suppressing laser frequency noise by about eight 439–447 (1975).
19
magnitudes in this way is part of what will make LISA’s John W. Armstrong, “Low-frequency gravitational wave searches using
gravitational wave detections possible in the first place.42 20
spacecraft Doppler tracking,” Living Rev. Relativ. 9, 1 (2006).
Deniz Soyuer, Lorenz Zwick, Daniel J. D’Orazio, and Prasenjit Saha,
“Searching for gravitational waves via Doppler tracking by future missions
ACKNOWLEDGMENTS to Uranus and Neptune,” Mon. Not. R. Astron. Soc. 503(1), L73–L79
(2021).
21
The author would like to thank Thomas M€uller and the This section profited greatly from the clear presentation of the basics of
anonymous referees for helpful comments on earlier pulsar timing arrays in Casey McGrath, “Gravitational wave timing resid-
versions of this text, and Norbert Wex for his help with the ual models for pulsar timing experiments,” Ph.D. thesis (University of
Wisconsin–Milwaukee, 2021).
section on pulsar timing arrays. 22
J. P. W. Verbiest et al., “Timing stability of millisecond pulsars and pros-
pects for gravitational-wave detection,” Mon. Not. R. Astron. Soc. 400(2),
AUTHOR DECLARATIONS 951–968 (2009).
23
Ronald Hellings and George Downs, “Upper limits on the isotropic gravi-
Conflict of Interest tational radiation background from pulsar timing analysis,” Astrophys. J.
265, L39–L42 (1983).
The author has no conflicts to disclose. 24
Joseph D. Romano and Bruce Allen, “Answers to frequently asked ques-
tions about the pulsar timing array Hellings and Downs curve,” Classical
DATA AVAILABILITY Quantum Gravity 41(17), 175008 (2024).
25
K. J. Lee, N. Wex, M. Kramer, B. W. Stappers, C. G. Bassa, G. H.
The data that support the findings of this study are avail- Janssen, R. Karuppusamy, and R. Smits, “Gravitational wave astronomy
able within the article. of single sources with a pulsar timing array,” Mon. Not. R. Astron. Soc.

07 July 2025 [Link]

414(4), 3251–3264 (2011).
26
There is another strong argument for having the default state be near dark,
a)
Electronical mail: poessel@[Link], ORCID: 0000-0003-0920-9532. but it goes beyond our simplified description: For reducing the quantum
1 noise (“photon shot noise”) mentioned in Sec. III A, it proves advanta-
Gabriella Agazie et al., “The NANOGrav 15 yr data set: Observations and
geous to increase the power of the laser light. A key technique for doing
timing of 68 millisecond pulsars,” Astrophys. J. Lett. 951, L9 (2023).
2 this is to tune the interferometer so that the default state at the photodetec-
EPTA Collaboration and InPTA Collaboration (J. Antoniadis et al.), “The
tor is near-dark. On its own, this would mean almost all of the laser light
second data release from the European Pulsar Timing Array III. Search for
would leave the Michelson interferometer at the other end, in the direction
gravitational wave signals,” A&A 678, A50 (2023).
3
Daniel J. Reardon et al., “Search for an isotropic gravitational-wave back- of the light source. Adding a semi-transparent “power recycling mirror” in
ground with the Parkes Pulsar Timing Array,” Astrophys. J. Lett. 951, L6 front of the light source, which reflects most of the returning light back
(2023). into the interferometer, significantly increases the amount of laser light
4
Heng Xu et al., “Searching for the nano-hertz stochastic gravitational that is inside the detector at each moment in time.
27
wave background with the Chinese Pulsar Timing Array data release I,” Reality is more complicated: The output signal is fed to actuators that
Res. Astron. Astrophys 23, 075024 (2023). actively compensate for a differential arm length change. That part of the
5
ESA, see [Link] gravitational wave that is slow enough for those actuators to keep up will
the_ripples_of_spacetime_LISA_gets_go-ahead for “Capturing the Ripples of be in the control signal, while the part that is too fast for the actuators will
Spacetime: LISA Gets Go-Ahead” (accessed January 25, 2024). be in the signal from the near-dark photodetector.
28
6
Benjamin Farr, GionMatthias Schelbert, and Laura Trouille, Markus P€ ossel (2023). “How an interferometric gravitational wave detec-
“Gravitational wave science in the high school classroom,” Am. J. Phys. tor works (animation),” Zenodo. [Link]
29
80(10), 898–904 (2012). Peter R. Saulson, “If light waves are stretched by gravitational waves, how
7
Rahul K. Choudhary et al., “Can a short intervention focused on gravita- can we use light as a ruler to detect gravitational waves?,” Am. J. Phys.
tional waves and quantum physics improve students’ understanding and 65(6), 501–505 (1997).
30
attitude?,” Phys. Educ. 53(6), 065020 (2018). Valerio Faraoni, “A common misconception about LIGO detectors of
8 gravitational waves.” Gen. Relativ. Gravitation 39, 677–684 (2007).
Joris van Heijningen, “Where do gravitational waves come from and how
31
can we detect more?,” in Magdalena Kersting and David Blair, Teaching B. Willke et al., “The GEO 600 gravitational wave detector,” Classical
Einsteinian Physics in Schools. An Esssential Guide for Teachers in Training Quantum Gravity 19(7), 1377–1387 (2002).
32
and Practice (Routledge, London and New York, 2021), Chap. 13. Benjamin P. Abbott et al., “Prospects for observing and localizing
9 gravitational-wave transients with Advanced LIGO, Advanced Virgo and
Gary W. Spetz, “Detection of gravity waves,” Phys. Teach. 22(5),
282–287 (1984). KAGRA,” Living Rev. Relativ. 21, 3 (2018).
10 33
E.g., Chap. 4 in Barbara Ryden and Bradley M. Peterson, Foundations of LISA CDF Team, see [Link]
Astrophysics (Cambridge U. P., Cambridge, 2021). 1567260318050-LISA_CDF_IFP_2017.pdf for “Internal Final
11
Sec. 3 in The LIGO Scientific Collaboration (J. Aasi et al.), “Advanced Presentation” (accessed May 5, 2017).
34
LIGO,” Classical Quantum Gravity 32(7), 074001 (2015). Peter R. Saulson, Fundamentals of Interferometric Gravitational Wave
12
J€oran Bauchrowitz, Tobias Westphal, and Roman Schnabel, “A graphical Detectors (World Scientific, Singapore, 2017).
35
description of optical parametric generation of squeezed states of light,” Adrian Melissinos and Ashok Das, “The response of laser interferometers
Am. J. Phys. 81(10), 767–771 (2013). to a gravitational wave,” Am. J. Phys. 78(11), 1160–1164 (2010).
13 36
E.g., Chap. 29 in Richard P. Feynman, Robert B. Leighton, and Matthew Matteo L. Ruggiero, “Gravitational waves physics using Fermi coordi-
Sands, Lectures on Physics (Addison-Wesley, Reading, MA, 1963), Vol. I. nates: A new teaching perspective,” Am. J. Phys. 89(6), 639–646 (2021).

509 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 509
37 40
F. I. Cooperstock and Valerio Faraoni, “Laser-interferometric detectors of Giuseppe Congedo, “Detection principle of gravitational wave detectors,”
gravitational waves,” Classical Quantum Gravity 10(6), 1189–1199 Int. J. Mod. Phys. D 26(05), 1741022 (2017).
41
(1993). Jo~ao C. Lobato, Isabela S. Matos, Lucas T. Santana, Ribamar R. R. Reis,
38
Lee Samuel Finn, “Response of interferometric gravitational wave and Maurıcio O. Calv~ao, “Influence of gravitational waves upon light in
detectors,” Phys. Rev. D 79(2), 022002 (2009). the Minkowski background: From null geodesics to interferometry,” Phys.
39
Michael J. Koop and Lee Samuel Finn, “Physical response of light- Rev. D 104(2), 024024 (2021).
42
time gravitational wave detectors,” Phys. Rev. D 90(6), 062002 Massimo Tinto and Sanjeev V. Dhurandar, “Time-delay interferometry,”
(2014). Liv. Rev. Relativ. 8, 1–54 (2005).

07 July 2025 [Link]

510 Am. J. Phys., Vol. 93, No. 6, June 2025 Markus P€

ossel 510