# Measuring possible one-way anistropy of light speed

Tags:
1. Mar 18, 2015

### seb7

wiki says "Experiments that attempted to directly probe the one-way speed of light independent of synchronization have been proposed, but none has succeeded in doing so"

Here's a proposed experiment which I could not find any evidence of this being performed before..

Central light source pulsing opposite directions at highly regular intervals. Two detectors, one either side at equal distance. Both detectors measuring 'jitter', ie. by looking at the timing between the pulses the detectors can determining a minimum and maximum value difference. These differences can be mapped while turning the device through 360 degrees, thus would very quickly highlight any differences in light speed in certain directions.

special relativity says light is 'independent of the state of motion', thus this experiment should be able measure movement? And if so, what movement would it detect say on the space station? Speed around the earth, speed around the sun, speed we are moving through the galaxy?

2. Mar 18, 2015

### PAllen

Your device would measure no motion, and would fail to detect the type of anisotropy that is consistent with SR (two speed invariant, and speed independent of source motion). If pulses are emitted at regular intervals (per a central clock), in a direction where light is 'slower', the pulse lengths and interval lengths will be smaller [precisely due to light being slower], preserving that any receiving clock's measurement of pulse duration and interval duration is unchanged as the device is slowly rotated. If you have something else in mind, you need to explain it much better.

Note, please be sure to phrase such proposals as questions about what is wrong, so you don't claim to contradict established physics (else this thread will be close quickly). Even so, there are so many threads on this that the topic is tiresome to most here, and may not draw much response. It can be a lot of work to find subtle flaws proposed experiments. This is similar to mathematicians being unwilling to locate the flaw in complex constructions for angle trisection, given that there is a proof of impossibility.

Last edited: Mar 19, 2015
3. Mar 19, 2015

### Mentz114

If your detector apparatus is moving inertially ( ie not firing rocket engines) then no differences would be detected,

4. Mar 19, 2015

### seb7

nope, only slowly turning as to determine straight line speed variation in different directions.

5. Mar 19, 2015

### harrylin

It think that it is likely that such set-ups have been used in the past for other purposes (and without surprises).
According to relativity, you cannot detect your speed that way; if you take the point of view that the speed of light is anisotropic relative to your set-up, then the clock synchronizations will be affected accordingly when turning your device, as from that point of view also the motions of the clocks are unequal. It is only possible to detect change in your velocity (acceleration) with such a setup.

6. Mar 19, 2015

### Ibix

You've got two detectors, so you either need a clock at each one or some communication link to a single comparator. In the first case you can't synchronise the clocks without making an assumption about the isotropy of the speed of light. In the second you are measuring a round-trip time (possibly with different signal types in different directions), not a one way speed.

7. Mar 19, 2015

### seb7

nope, using synchronized clocks isnt any good for looking at one-way light speed. Not timing source to detector, but looking for tiny changes in the beat, as to calculate a relative high and low value. Im trying to completely avoid round-trip time. Actually I dont even need two detectors really, but would help in comparing the jitter results in realtime.

8. Mar 19, 2015

### PAllen

You need to much better describe your proposed setup. As I understand it, it will detect nothing. It does not appear anyone understands what you have proposed. Perhaps start by explaining in detail what you mean by measure jitter, and upload a picture of your proposed experiment. Be as precise as you can. Your OP is completely imprecise and incomplete.

9. Mar 19, 2015

Staff Emeritus
I agree with PAllen, the setup is ill-defined, or at least ill-described. Nevertheless, it won't work. There are three things you need to know to predict the time that a clock will read when a light pulse hits it - the distance to the source, the speed of light along that path, and the clock synchronization convention. In doing the experiment, you measure two numbers: distance and time. In short, you have too few measurements to constrain too many variables: two equations in three unknowns.

Adding complexity is increasing the number of measurements, but each new measurement suffers from the same problem as the single measurement, so it doesn't matter how complicated a setup you use. There is simply not enough information to determine both the one-way speed of light and the clock synchronization convention. In the Einstein clock synchronization convention, where clocks are brought together, synchronized, and then slowly moved to their final position, the speed of light is isotropic.

10. Mar 19, 2015

### seb7

attached a sketch

#### Attached Files:

• ###### device.jpg
File size:
35.9 KB
Views:
154
11. Mar 19, 2015

### seb7

I should add that its not attempting to measure the speed of light, but to measure the (one-way) differences depending on direction.

12. Mar 19, 2015

### Mentz114

I don't understand what this is supposed to measure. What are 'one-way' differences ? I have to say I'm really impressed. A flashing light - who'd a thunk it.

13. Mar 19, 2015

### harrylin

I supposed -and still do- that you attach a clock to each detector. Is that so?
Then I correctly understood your set-up when I gave my comment in post #5.

14. Mar 19, 2015

### seb7

yes, it detects changes. My question is whether it detects the direction of current velocity. (not just acceleration).

Sorry over simplification; but yes, this flashing light would be a regular pulse that would likely need an atomic clock or similar mechanism for the amount of accuracy required for this to work at all.

I suppose it would only measure acceleration, since the device being central to the observation would only ever see light traveling at the same speeds in both directions.

15. Mar 19, 2015

Staff Emeritus
You still have one more unknown than you have constraints. This is still impossible.

16. Mar 19, 2015

### harrylin

That would break the relativity principle; however you agree to apply Special Relativity (SR). SR is based on not one but two assumptions which at first sight are contradictory. As applied to your set-up:

1. You can not detect your "true speed" (if that even exists) relative to the light rays.
2. Light propagates like a wave: the speed of a wave is independent of the speed of the source.

How much do you know of the Lorentz transformations?

17. Mar 19, 2015

### PAllen

Then, I understood it as well, and my comments so far remain. Your device would not measure any anisotropy even there 'really is' anisotropy of the type allowed by constraints of two way light speed measurement invariance, and no affect of speed of emitter. Further, it would not measure any motion as long as there were no acceleration.

18. Mar 19, 2015

### Staff: Mentor

19. Mar 19, 2015

### Staff: Mentor

You started this thread with "Here's a proposed experiment which I could not find any evidence of this being performed before" and then described a thought experiment that (despite the misleading thread title, which I have fixed) would not measure the speed of light, but instead would seek to measure whether it changed in different directions.

A flashing light and an atomic clock is impractical for this purpose, and the atomic clock introduces extraneous concerns about clock synchronization. However, what you're really trying to do is to superimpose the signal generated from the central source on top of a signal generated by an identically constructed local oscillator at selected points on the shell, and looking for changes in the relative phase (the "jitter" that you mention) when the apparatus is slowly rotated while at rest in an inertial frame.

Special relativity predicts that no such effect will be seen, and HarryLin was understating the case when he said above that

20. Mar 19, 2015

### Mentz114

You quoted me in this reply. I was being sarcastic, which always leads to misunderstanding.

I think the idea is childish, frankly. And to think that no one has thought of it is naive at least. The poster should go away and do some serious studying and stop reading Wiki pages.

I think I've made myself clear now.

21. Mar 19, 2015

### Ibix

I had misunderstood the setup. I thought he was going to compare the pulses received at the two arms. This can either be done by "comparing the video tapes" or "side-by-side live CCTV", which is where my post was coming from.

All the rigs have variants on the same problem, of course.

22. Mar 19, 2015

### Staff: Mentor

@seb7 You should study the Edwards synchronization convention. It has isotropic 2 way speed of light but anisotropic one way speed of light. You can transform from a Lorentz frame to such a frame. Since it is just a coordinate transform, none of the invariants differ.

23. Mar 20, 2015

### DrGreg

I think I understand the thinking behind this experiment. And I also understand the flaw in its design.

We have an inertial light source, equipped with a clock, which emits pulses of light at regular intervals according to its clock. In fact we need only one receiver, equipped with its own clock, that moves very slowly in a circle around the emitter. Both the emitter and receiver plot graphs of pulse count versus their own elapsed (proper) time.

The emitter's graph will be a straight line. With isotropic one-way light speed we expect the receiver's graph to be a straight line, too. But with anisotropic one-way light speed, the experiment designer expects to get a curve. On some parts of the circle the pulses should arrive sooner than expected, and at the diametrically opposite points, later than expected. The receiver's graph would deviate from a straight line by a roughly sinusoidal perturbation, the designer argues.

What's wrong with this?

There is a hidden assumption that the motion of the receiver's clock does not invalidate the measurements. We know that clocks in relative motion may lose their synchronisation to each other. That's the well known "twin paradox". But, surely, if the receiver moves slowly enough, these effects become negligible? It's true that the time dilation factor ($\gamma$) decreases towards 1 as the speed decreases to zero. But the time taken to complete a circle gets longer (in fact, towards infinity!) as the speed decreases, and this increases the dilation effects. The decrease and increase tend to cancel each other out so that the relativistic consequences of motion do not become negligible.

Let's suppose there are a number of inertial clocks all distributed around the circle, at rest relative to the central emitter. As the receiver moves round the circle, each of the fixed clocks gets synchronised to the moving clock. The experiment then becomes equivalent to using these fixed clocks to measure the one-way speed of light. This method of synchronisation is called "slow clock transport". It can be proved that synchronisation by slow clock transport gives exactly the same result as Einstein synchronisation (which, by definition, guarantees isotropic one-way speed of light). (See discussion, and references, at One-way speed of light.)

The conclusion is that this experiment can't detect any one-way anisotropy (assuming two-way isotropy is already established).

24. Apr 6, 2015

### seb7

sorry, I dont think anyone understood my experiment.

I keep reading how there hasnt been an experiment that can measure one way speed of light, but the experiment I propose though wouldnt measure that actual speed, it can compare light speed from one direction against another direction, and show you the offset. (if any)

Physics says the resulting speed is the same in all directions. But I've yet to see any an experiment that directly demonstrates this other than my proposal.

Someone asked (in this thread) if I knew about 'Lorentz transformations'... yes. It seems the formula just treats time as a 4th dimension travelling at the speed of light, its seems like an over complication of the pythagoras theorem to me.

25. Apr 6, 2015

### Ibix

I think you are confusing the Lorentz transforms with the interval, which is exactly Pythagoras in a non-Euclidean four-dimensional space.

That's not directly relevant to this answer, though.

You are imagining a light source emitting regular pulses. The pulses spread out like ripples on a pond. If the speed of light is isotropic, those ripples are circular, and a clock on the end of an arm travelling in circles around the source ticks in perfect synchronisation with the arrival of each ripple. However, if the speed of light is not isotropic the ripples will be eliptical (or non-circular, at least). The clock circling the source will drift in and out of sync with the ripples. Right?

Dr Greg explained why it doesn't work like that. Your moving clock's tick rate will vary by exactly the right amount to mask the variation in pulse arrival time. What you are doing is called slow clock transport, and is exactly equivalent to Einstein synchronisation, which is just directly synching the moving clock to the pulse arrival time.

The speed of light is not just the speed at which light travels. It is a property of space-time. Anything in space-time gets adjusted the same way if that property changes, so one way anisotropy measures always give null results because your equipment always changes the same way what you are trying to measure changes.