Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

One way speed of light measurement proposal

  1. Mar 7, 2012 #1
    Hello all,

    Having read many threads and posts about light, it’s speed c, SR etc, the particular issue of not being able to measure the one way speed of light has always stood out and kept me very interested, so, after pondering about it for a while, I would like to propose an experimental setup in order to measure the one way speed of light.

    The attached diagram shows the simple setup which has a source, two mirrors and one clock;

    - Distance SC = distance SM2 = 10 meters
    - mirrors M1 and M2 are positioned at a 90 degree angle from each other

    Am I right in saying that for an observer at rest with the test setup;

    1- the light pulse will start the clock (t1 = 0) and reflect from mirror M2 simultaneously.

    2- the reflected pulse from M2 will stop the clock at t2 and, via Pythagoras, a calculation of the one way speed of light is given by the equation

    SqrRoot ((M1M2)^2 + (M1C)^2) / (t2-t1) = SOL

    Using a rotating gantry, this experimental setup could be repeated in any starting direction to gather data for additional validation.

    Can this be a valid measurement of the one way speed of light?

    Thank you for your comments and best regards,


    Attached Files:

  2. jcsd
  3. Mar 7, 2012 #2
    Only if you assume the speed of light is the same in perpendicular directions - but if you assume that, then the one-way speed equals the two-way speed by definition.

    Regardless of whether you're trying to measure the one-way or the two-way speed of light (or anything else for that matter), you need to establish your basis of measurement, meaning a suitable measure of space and time intervals. Without establishing such a basis, you obviously can't measure anything. Once you have established such a basis, measuring the speed (one-way or two-way) is straightforward.
  4. Mar 7, 2012 #3
    Just to double-check that I’ve read your diagram correctly, your idea is that the light pulse that travels from S is split by the half-silvered mirror, M1. The part that is transmitted through M1 starts the clock C when it hits it (initially set at t1=0). The light reflected from M1 towards M2 then reflects off M2 towards the clock, stopping the clock when it arrives (t2).

    If this is the case, your calculation is correct.

    Regarding using a rotating gantry, what hypothesis are you interested in confirming with the additional data? Is it anything to do with the fact that many explanations in relativity employ the use of light reflected back to its source position?

    (An alternative would be to use no mirrors, and two synchronised clocks, at the start and the end of the light path. Of course, one of the methods suggested for synchronising clocks involves reflecting light between them, so perhaps this isn't what you're after!)
  5. Mar 10, 2012 #4
    You are still using two paths.

    A strict 'one way' experiment should only use one, and, it wouldn't really work, would it??

    I'm of two minds there as 'gravitational time dilations' should be everywhere in its path? Even though light is a constant it's so locally, but the observer will define its speed 'globally', over the whole path. In the two way experiment where it is reflected it should equal out, but in a one way experiment? And if you're a true believer in symmetries, as me is, joined by their 'complementary' Lorentz FitzGerald contractions.


    If someone has a good answer to that one I would be grateful.
    It was a sweet idea with entanglements :)

    The more I think of it the more confused I feel ::))
  6. Mar 12, 2012 #5
    Hello all,

    Samshorn, you wrote;

    Indeed, this is the assertion I’m making here; that the light reaches both C and M1 simultaneously because the speed of light is the same in all directions, and I think this setup makes it possible to validate.

    That it is equal to the two way speed by defenition is not an issue here, regardless of any predetermined parameters or assumptions, I think doing this simple experiment could be interesting, just to gather the data, just to see if the T2-T1 difference corresponds to the expected value in order to get c... I beleive it will.

    Goodison Lad, you wrote;
    Well, no hypothesis in particular but rather compare all the different values of T2-T1 when the setup is rotated... I beleive there will be no discrepencies.

    Yoron, you wrote;
    Well, if this is using two paths, then I’m actually using three... the M2-C path is another one altogether.

    Even if this is not a true one way measurement or validation, it still is only using one clock, no need for any synchronisation, and the distances between the instruments can be very accurately measured, or is 10m East not the same as 10m South ?

    So, either T2-T1 will yield the current value of c or it won’t, it’s as simple as that... I beleive it will, in any and all directions.

  7. Mar 15, 2012 #6
    one way speed is two clocks, spatialy seperated, and where simultanity creeps in.

    two way speed is one clock.

    I agree with Yoron that this is a there & back measure of c, and there being only one clock...
  8. Mar 15, 2012 #7
    Hello again, ValenceE

    If you moved the clock closer to M1 so that it was right next to it (adjusting the orientation of M2 appropriately), then the clock would start at the same time as the light from S was transmitted through/reflected from M1, and stop again on reception of the reflection from M2.

    Effectively, then, you would be measuring the round-trip time for the beam that went from M1 to M2 and back again. I think, then, that this set up is essentially the same as your original proposal, and is a round-trip measurement.
  9. Mar 15, 2012 #8
    Afterthought: you would, of course, be justified in saying that I'm talking about a different experiment, because your original set-up was an ingenious way of basically clocking the travel time from M2 to C. This to me weakens the point I made in my last post!

    So, if it is assumed that the velocity of light is independent of direction (as suggested by Samshorn you would have to), I'm starting to think it is a one-way measurement. Every piece of apparatus is static, so there's only one frame of reference involved.

    Fickle, aren't I?
  10. Mar 15, 2012 #9
    No it doesn't. It's well known that, if the speed of light depends on direction, it must be of the form C(theta) = c/[1 + k cos(theta)] for some constant k between 0 and 1, where theta is the angle of the direction of the light ray relative to some fixed reference. For example, let's take the line from your source to your clock to be the theta=0 direction, so the speed of light is c/(1+k) in that direction and c/(1-k) in the opposite direction. It follows that the speed in the perpendicular direction (from M1 to M2 in your setup) is simply c, and the speed in the "hypotenuse" direction is c/[1 + k/sqrt(2)]. Plug all these speeds into the setup, and you find that the elapsed time between the pulses reaching the clock is always exactly L sqrt(2)/c, regardless of the value of k. So you get the same result, regardless of whether c is the same in all directions or not (i.e., regardless of whether or not you have chosen a coordinate system in which c is the same in all directions).

    Indeed it will, but it will do so regardless of whether or not the speed of light is the same in all directions. In fact, it's easy to show that any complicated set of light paths you can imagine (not just a simple triangle) will all give the same results, provided only that the speed of light has the directional dependence given by the ellipsoidal form noted above. You can read all about this in any good book on relativity.
  11. Mar 16, 2012 #10
    You have a really god point in that you're using one clock. Maybe? In SR naturally. I'ts quite nice.
    The thing is, every time I think of a 'one way experiment' I remember NIST, and get a headache :) But that is GR.
  12. Mar 18, 2012 #11
    This experiment has really got me – I’m arguing with myself over it in my sleep!

    It occurs to me that, in using your setup to determine the travel time of the light from M2 to the clock, the implicit assumption is that the reflected light reaches M2 simultaneously with the transmitted light reaching the clock i.e. time at M2 is synchronised with time at the clock.

    In ordinary two-way measurements, no such assumption is necessary, since the light leaves and returns to the same clock.

    Since the measurement of the one-way speed of light requires two different locations, we need to know something about the synchronisation of time at those two points in order to determine the flight time of the light between them.

    So the question becomes: how would we know, without assuming it, that time at M2 is synchronised with time at the clock?
  13. Mar 18, 2012 #12


    User Avatar
    Science Advisor
    Gold Member

    There is no question but that this experiment will give a result that is consistent with Einstein's postulate that the one-way speed of light is equal to the two-way speed of light, in other words, c. If it didn't, it would prove Einstein's postulate to be wrong.

    But this experiment and all others like it are designed to measure the ether wind, or the rest state of the ether, which is the same thing. If super stable clocks and super accurate rulers and super rigid structures were available at the time of MMX, they could have performed experiments on the one-way speed of light instead of just on the two-way speed of light, but they would have come to the same conclusion, that they couldn't measure any ether wind, and that would have thrown them into exactly the same theoretical quandary that they were thrown into with the null results from their two-way measurements and they would have come to the same conclusions, based on their persistent belief in an absolute ether rest state and absolute time and absolute space, that their super stable clocks and super accurate rulers and super rigid structures were in fact not so super stable, accurate and rigid as they thought, not that time and space were relative.

    The point is, if you start with the concept of an absolute ether rest state in which light travels at c, then you will interpret the result of this experiment as not really measuring the one-way speed of light, but rather that your rulers contract along the direction of motion through the ether and that your clock is time dilated and your structure deforms as you rotate it to always yield the same measurement for the one-way speed of light.

    Even in Special Relativity, with Einstein's convention for establishing a Frame of Reference, which is identical to the concept of being at rest in the absolute ether rest state, another identical experiment performed with a relative motion in that FoR is interpreted to not be measuring the defined one-way speed of light, but rather to be subjected to the same length contraction, time dilation, and deformation that would explain the null result without Einstein's convention.

    So the bottom line is, there is no experiment that can determine the actual one-way speed of light, apart from a previous assumption about what that one-way speed of light is. Or to put it another way, as Einstein said, apart from defining the one-way speed of light, we can't know what it is, that is, we can't measure it.
  14. Mar 18, 2012 #13
    Just to add another perspective to this, while it is certainly true that the one-way speed of light cannot be defined independent of simultaneity convention, there are some simultaneity conventions which unlike Einstein synchronization do not define the one-way speed of light in advance, but rather the one-way speed of light under such a convention has to be determined experimentally (and such an experiment can constitute a test of SR). See this long discussion.
  15. Mar 18, 2012 #14


    User Avatar
    Science Advisor
    Gold Member

    Does ValenceE's "One way speed of light measurement proposal" qualify as an experimental determination of the one-way speed of light since it did not define that speed in advance?
  16. Mar 18, 2012 #15
    No, of course not. You can't measure the one-way speed of light, tautologically or otherwise, with only one clock. You and I are in complete agreement for the purposes of this thread.
  17. Mar 19, 2012 #16


    User Avatar
    Science Advisor
    Gold Member

    But you can with two clocks?
  18. Mar 19, 2012 #17
    Hello ghwellsjr and lugita15,

    Thank you both for your comments…

    I do agree that in this setup, we don’t know the SOL for paths M1-C and M1-M2, we don’t know if they are equal and, if not, which is ahead or behind, we only know that the paths are of equal lengths and that the M2-C length is sqrt(200), the only measured value being T2-T1.

    Let’s say you do this experiment 360 times, rotate the apparatus 1 degree at a time, and record the result of [sqrt(200)/ (T2-T1)], I’d really like to know if it would match the accepted value of c… my belief is that it would on all 360 readings.

    Again, there is only one clock, no to-from measurements, nothing assumed…

    So I ask you, given what we know about SR;

    - what can we make of the results for the ones that do match?, those that don’t ?
    - what if they match for all 360 readings ?
    - what can we make of the results if none match?



    PS: As far as ether is concerned, I don’t believe in an absolute ether rest state but I have ideas about a dynamic one, but that is another subject altogether, although related…
  19. Mar 19, 2012 #18
    You do realize that all your questions were answered in post #9, right?
  20. Mar 19, 2012 #19
    hello Samshorn,

    actually, no I don't... I wanted to respond to your post #9 earlier but got sidetracked.

    You say that;

    "it must be of the form C(theta) = c/[1 + k cos(theta)] for some constant k between 0 and


    "...so the speed of light is c/(1+k) in that direction and c/(1-k) in the opposite direction."

    From this it follows that if K=1 then you get c/2 in one way wich to me seems awfully slow, and how about c/0 in the other way ? is it undefined or infinite?

    Sorry, but I'm just not comfortable with those values...

    regards, QE

    edit: sorry again, but I must leave now... will be back later today, thx
  21. Mar 19, 2012 #20
    We've covered this territory pretty thoroughly, but to repeat you can synchronize two clocks according to slow transport, and then if you do a measurement of the one-way speed of light with respect to these clocks, you cannot predict the result in advance, i.e. without knowing what universe you're in. Whereas if you synchronize them with respect to Einstein synchronization, you immediately know that the one-way speed of light with respect to these clocks will be isotropic, even if you don't know what universe you're in or what the laws of physics are.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook