Could the following experiment help is to compare the one way and the two way speed of light.

We sent a lightpulse from a clock in A. This lightpulse is reflected in B but also activates a lightpulse in B. Both pulses travel to A. We now know the distance between A.and B and we know if there is a difference in travel time between the one way speed from B and the returning speed from A to B.

Moreover if there is a difference we know the difference in time.

Why can't this work.

  • $\begingroup$ I'm not sure what is happening here. You talk of a "returning speed" from A to B, but I did not see anything metnioning a "return" pulse from A to B. The only pulse sent was the initial one. $\endgroup$
    – Cort Ammon
    Commented Dec 14, 2019 at 4:43
  • $\begingroup$ How do you think light could have different speeds? $\endgroup$ Commented Dec 14, 2019 at 5:28
  • $\begingroup$ @Adrian Well, we don't think light has different speeds (locally, in a vacuum), but there's a little problem: nobody has thought of an experiment that measures the one-way speed of light independent of a clock-synchronization convention. See en.wikipedia.org/wiki/Einstein_synchronisation en.wikipedia.org/wiki/One-way_speed_of_light and pitt.edu/~jdnorton/teaching/HPS_0410/chapters/… $\endgroup$
    – PM 2Ring
    Commented Dec 14, 2019 at 12:51
  • $\begingroup$ We also remember that one way speed of light is anisotropic relatively to a rim of rotating ring (Sagnac effect) hence it is anisotropic relatively to any laboratory on Earth surface including that of Michelson-Morley. Surprisingly, prof. Norton completely forgot to mention this fact in his long essay. Sure, proposed by OP experiment is not able to measure times of propagation of light pulse in different directions without certain convention as how to synchronise clocks in A and B. $\endgroup$
    – user139020
    Commented Dec 14, 2019 at 13:52

2 Answers 2


My former mentor was obsessed with this question at first saying there's no way to prove the anisotropy of light (one-way vs reflected 2-way) until Don Lincoln responded to just slowly separate two sync'd atomic clocks and fire light from one to the other and measure the delay. The problem is moving clocks apart causes them to unsync but this is not a problem for a single clock that measures its own reflected beam. So moving them slowly apart introduces an error which can be accounted for using relativity.

Einstein's clock sync method uses light pulses to sync clocks and then you use those light sync'd clocks to measure the speed of light. The result is dependent on the assumption so it's a circular argument and is deemed untrustworthy.

In Einstein's day (1905) they didn't even yet know about atoms let alone atomic clocks so his clock sync method was all he had. Today we can depend on the universal accuracy of atomic clocks to free us from his method (which is ingrained into the derivation of the equations of relativity).

My solution is to move the clocks apart at a slow constant velocity and fire light from one to the other at a pre-agreed proper time on each clock without stopping. The light will meet where they started and you could use an inteferometer to measure any discrepancy in the one-way light speed. When both clocks move at constant velocity, they both tick at the same proper time rate. It's a simultaneity (relative to their common starting point) that is independent of distance separation. If you stop the clocks, you invoke the twin paradox and a syncing problem between them. I recently wrote something here that could add further clarification but it's in limbo. I'm not sure if the experts disapprove or if it's still confusing.

  • $\begingroup$ If the speed of light is different in different directions, then time also does not dilate uniformly in all directions. $\endgroup$ Commented Mar 12, 2021 at 16:49
  • $\begingroup$ Yes that's why using a center stationary clock between the other two that are separating would check for any difference in c due to direction as this video brings up. youtube.com/… $\endgroup$
    – ralfcis
    Commented Apr 4, 2021 at 20:16
  • $\begingroup$ That suffers from exactly the same problem though. You can't detect the clock skew in that manner without already assuming isotropy. $\endgroup$ Commented Apr 5, 2021 at 17:46
  • 1
    $\begingroup$ You really should post this as a question rather than an answer on somebody else's question. But: the problem with your method is making sure that the 2 moving clock have the same speed. In order to verify a speed, you need to measure the distance between two points, and the time it takes to travel between those two points. For measuring the time it takes, you need two other clocks, one at one of the points, and one at the other. These two other clocks need to be synchronised with each other, likely with the Einstein synchronisation method.... $\endgroup$
    – fishinear
    Commented Apr 23, 2021 at 14:42
  • 1
    $\begingroup$ .... If the speed of light is different in two directions, then that synchronisation method will cause the two clocks to be out-of-sync (compared to an omni-directional uniform light speed). The end result is that, while you think that they have the same speed, they really haven't, and their time-dilation will be different. Everything conspires together for the end result that the light still arrives at the mid-point at the same time, in spite of the different light-speeds in different directions. $\endgroup$
    – fishinear
    Commented Apr 23, 2021 at 14:47

I read your question as following:

  • The speed of light travelling from $A$ to $B$ might be different from the inverse travelling direction, from $B$ to $A$. Thus, you suspect that the speed of light depends on the traveling direction. This idea is analog to the Michelson-Morley hypothesis, were light travels inside the ether. The negative result lead Einstein to his famous conjecture: The speed of light is constant and does not depend on the reference system.
  • Instead of taking a clever set-up (interferometer), where we only need one accurate clock and most of the measurement problems by measuring in two states (rotating the apparatus), you suggest that we use two accurate clocks: The first one at position $B$ -- so that we "know" the time the light takes to travel from $A$ to $B$ -- and a second one at position $A$ -- in order to "know" the travelling time from $B$ to $A$.

Please let us know, if these statements are correct. If they are correct, here is my answer: In principle your set-up should work, but ...

  • I don't understand, why we know the travel time from $A$ to $B$ in your set-up. How does the clock at $B$ know the starting time?
  • I believe you want to make sure that the reflected light and the at $B$ generated light has only a "small" time delay. While this will be hard to do (depending on your definition of "small" time delay) it is also unnecessary: If you are able to measure the time from $A$ to $B$ why don't you just reverse the experiment and measure the time from $B$ to $A$.

The Michelson-Morley experiment uses a very clever set-up, because it eliminates all kind of systematic errors. Their key idea was, that instead on measuring time, they measure frequency -- the interference fringes can be counted. This philosophy/advice is followed in modern physics whenever we need to obtain accurate timings.

Does this answer your question?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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