This question about the speed of light prompted my own question. In the linked question it is asked if there is experimental proof that the speed of gravity equals the speed of light. I was surprised not to see the LIGO measurements mentioned.
The experiment uncovered the arrival of a spacetime distortion coming from fast-spinning binary systems of black holes or neutron stars. Due to LIGO's extensive Nature (there is one observatory in Livingston and one in Hanford) it seems that upon arrival the gravitational wave (if it hits the Earth at a sharp angle) will hit one of both observatories first (which one depends obviously from the origin of the wave). So it should be possible to measure the speed of gravity. Or, at least, to measure if the speed is finite (or not).
Has this been done?


4 Answers 4


Yes. In principle, the speed of gravitational waves can be measured using the data of LIGO. In fact, using a Bayesian approach, the first measurement of the speed of gravitational waves using time delay among the GW detectors was suggested/performed by Cornish, Blas and Nardini. By applying the Bayesian method, they found that the speed of gravitational waves is constrained to 90% confidence interval between $0.55c$ and $1.42c$ by use of the data of binary black hole mergers GW150914, GW151226, and GW170104.

After that, a more precise measurement of the speed of gravitational waves was performed by the measurement of the time delay between GW and electromagnetic observations of the same astrophysical source, as @Andrew nicely mentioned, by use of a binary neutron star inspiral GW170817. They found the speed of gravitational wave signal is the same as the speed of the gamma rays to approximately one part in $10^{15}$. Note that this study is primarily based on the difference between the speed of gravity and the speed of light.

Recently, a new method has been introduced using a geographically separated network of detectors. As the authors mentioned, while this method is far less precise, it provides an independent measurement of the speed of gravitational waves by combining ten binary black hole events and the binary neutron star event from the first and second observing runs of Advanced LIGO and Advanced Virgo. By combining the measurements of LIGO and Virgo, and assuming isotropic propagation, the authors have constrained the speed of gravitational waves to ($0.97c$, $1.01c$) which is within 3% of the speed of light in a vacuum.

In my opinion, the best study is the second one (that @Andrew nicely mentioned), in which multiple measurements can be measured to produce a more accurate result, but the later (the third study) has its scientific significance. This is because the later method is an independent method of directly measuring the speed of gravity which is based solely on GW observations and so not reliant on multi-messenger observations, as the authors mentioned.

Besides these achievements, there are other interesting results that one can extract from LIGO's data. For example, observations of LIGO have constrained a lower bound on the graviton Compton wavelength as

$${{\lambda _{{\rm{graviton}}}} > 1.6 \times {{10}^{13}}{\rm{km}}},$$

which is really interesting. In fact, assuming that gravitons are dispersed in vacuum like massive particles, i.e. ${\lambda _{graviton}} = \frac{h}{{{m_{graviton}}\,c}}$, one can find an upper bound for graviton's mass as ${{m_{{\rm{graviton}}}} \le 7.7 \times {{10}^{ - 23}}eV/{c^2} \sim {{10}^{ - 38}}g}$, which is extremely small, beyond the technology of our detectors.

  • $\begingroup$ It strikes me (as a layman observer) that the confidence intervals extend above the speed of light. Is there any scientifically compelling reason why they are allowing for the possibility of something moving faster than light and not just truncating at 1c? $\endgroup$
    – Bamboo
    Mar 24, 2021 at 6:48
  • 1
    $\begingroup$ @Phill how would you know that the speed of gravity cannot exceed the speed of light? Of course we know for everthing we have measured so far, but these are things we never measured before. Also in physics experiment always trumps theory. $\endgroup$
    – lalala
    Mar 24, 2021 at 9:58
  • $\begingroup$ SGB, aren't there difficulties here because of the magnitude of the expected speeds and complications dealing with the impossibility to rigorously define simultaneous events? For instance, and because of these things, it's not possible to measure the speed of light in a one-directional transit and only round trips can be considered. Could these difficulties have something to do with the need for a Bayesian method? $\endgroup$
    – ttonon
    Mar 26, 2021 at 18:26

Yes, the multi-messenger observation of GW170817 (a binary neutron star merger) allows for a measurement of the speed of gravitational waves. One can compare the arrival times of the gamma ray burst (which travels from the source to Earth at the speed of light), with the arrival time of the gravitational waves (which of course travel at the speed of gravitational waves).

The test is described in detail in this paper published in The Astrophysical Journal Letters:

Abbott et al. (2017), Gravitational Waves and Gamma-Rays from a Binary Neutron Star Merger: GW170817 and GRB 170817A, https://doi.org/10.3847/2041-8213/aa920c.

In particular, LIGO, Virgo, and the Fermi Gamma-ray Burst Monitor found that

\begin{equation} -3 \times 10^{-15} \leq \frac{\Delta v}{v_{\rm EM}} \leq +7 \times 10^{-16} \end{equation} where $\Delta v=v_{\rm GW}-v_{\rm EM}$ is the difference between the speed of gravitational waves and the speed of light, $v_{\rm GW}$ is the speed of gravitational waves, and $v_{\rm EM}$ is the speed of light.


The time delay between LIGO detections doesn’t so much verify the speed of gravity to equal $c$; rather, this fact is taken as a given to help determine the direction of origin of the wave. For example, if the two detectors register a blip at the same time, then the wave must have come from some direction perpendicular to the line connecting the two. A measurement of the delay restricts the possible originating directions, but only if the speed of gravity is known a priori (which LIGO can help to determine by other means, as discussed in another answer).

  • 3
    $\begingroup$ This is incorrect. See my answer, LIGO and Virgo have placed very strong constraints on the speed of gravitational waves. $\endgroup$
    – Andrew
    Mar 21, 2021 at 15:53
  • 1
    $\begingroup$ @Andrew But surely triangulation assumes that gravitational waves travel at the speed of light - and that assumption is then confirmed by the near-simultaneous detection of electromagnetic radiation from the expected direction. $\endgroup$
    – gandalf61
    Mar 21, 2021 at 16:05
  • 5
    $\begingroup$ @Andrew OP specifically asked about a speed of gravity measurement from the time delay between detections of the same gravitational wave by two displaced detectors. Your own answer is good, but it answers a different question. Namely, it discusses the experimental evidence supporting the a priori assumption of the speed of gravity that I mentioned above. My answer stands. $\endgroup$
    – Gilbert
    Mar 21, 2021 at 16:22
  • 3
    $\begingroup$ @Andrew I agree with Gilbert's narrower reading of the question. OP very clearly delineates a specific way to use the detectors, and this answer addresses that directly. Your broader reading provides very useful information, but downvoting valid answers for answering the question as posed isn't great. $\endgroup$ Mar 21, 2021 at 20:36
  • 2
    $\begingroup$ @EmilioPisanty Fair enough but it also isn't correct that LIGO provides no information on the speed of gravity, even in the way the you and Gilbert understand the OP to mean, as pointed out by SG8, using the formalism from Cornish et al journals.aps.org/prl/abstract/10.1103/PhysRevLett.119.161102. My main issue was with implying that LIGO and Virgo cannot provide any information about the speed of gravity, which is scientifically not correct. But this is resolved now in the answer and I have upvoted. $\endgroup$
    – Andrew
    Mar 21, 2021 at 23:48

Time differences in the detection of a gravitational wave event at two or even three detectors can only give an upper bound on the speed of gravitational waves. This is because there is a line of points that are equidistant from all three detectors. You would need to detect an event at at least four sites to establish the speed of gravitational waves from time differences in the gravitational wave detection alone. So far, only one event (GW170814) has been detected at three different sites, and no events have been detected at four sites.

However, this is not a major problem because, as mentioned in another answer, comparing arrival times of gravitational waves and electromagnetic waves from GM170817 has already provided very tight upper and lower bounds on the speed of gravitational waves.

  • 1
    $\begingroup$ "So far, only one event (GW170814) has been detected at three different sites" that is so early 2020 :). With the publication of the 2nd GW catalog GWTC-2, there now are around 50 confirmed GW detections. I cannot quickly find the exact number, but this must include more than a handful of triple detections, in which the two LIGO's and Virgo were all online and saw a signal with significant SNR. Note this only includes data up to the O3a science run, O3b is still to be published. $\endgroup$ Mar 23, 2021 at 20:56

Not the answer you're looking for? Browse other questions tagged or ask your own question.