How to measure the one way speed of light.

[harrylin]

What do you means is made?

Light is isotropic in ECI otherwise, you could not synch clocks in it. This is a well know fact in the mainstream.

That is presenting the facts upside-down! At the risk of SRT being deleted by "mainstream", here's the "principle of the constancy of the velocity of light" as reiterated by Einstein in 1907:

"We [...] assume that the clocks can be adjusted in such a way that
the propagation velocity of every light ray in vacuum - measured by
means of these clocks - becomes everywhere equal to a universal
constant c, provided that the coordinate system is not accelerated.

- http://www.soso.ch/wissen/hist/SRT/E-1907.pdf

Harald

 

[Q-reeus]

The relations for the transformations are fundamentally fixed in variable form regardless of the kinematic theory for how the universe operates, whether it be SR or LET or anything else. Let's say we are stationary in a ship within our own frame and an identical ship passes us in the opposite direction along the x axis, identical in the respect that they measure the same proper length d of their ship as we do of ours. Both ships synchronize the clocks at the front of their ships to T=0 upon passing, when the fronts coincide, which is the origin of each frame. According to our frame, the other ship's clocks are ticking slower by a factor of z, their rulers and ship are length contracted by a factor of L, and they have synchronized so that to us, the clock at the rear of their ship reads a greater time than the front clock by tl, so the other frame is synchronized with an additional time of tl per length L d. An event then occurs according to our frame at coordinates t, x. According to the other frame, the event would occur at x - v t, but their rulers are also contracted, so they measure

x' = (x - v t) / L

The event occurs at time t to us, so when the clock at the front of the other ship reads z t, but the simultaneity difference adds a time tl per length L d in the negative direction also, so the time upon a clock that coincides with the event in the other frame will read

t' = z t - (tl / (L d)) (x - v t)

Those are the fundamental forms of the transformations which would be used regardless of the kinematic theory, so variable with LET. Of course with SR, we have the symmetry principle, whereby z = L = sqrt(1 - (v/c)^2) = 1 / y and tl = z L d v / (c^2 - v^2) = d v / c^2, which gives

x' = y (x - v t)

t' = t / y - (y v / c^2) (x - v t)

= y (t / y^2 - (v / c^2) (x - v t))

= y (t (1 - (v/c)^2) - v x / c^2 + (v/c)^2 t)

= y (t - v x / c^2)

Understand what you're getting at I think - it's only the inter-frame measure product x't' that's 'invariant' for a given relative frame velocity, and either factor x' or t' could vary arbitrarily within that, so L can vary arbitrarily. Problem I see here is that SR Doppler shift is known to high precision - that fixes the t' part (the t'= y (t - v x / c^2)), thus forcing....

 

[Q-reeus]

It's an essential point of SRT that because clocks can be synchronized according to our wish, for our convenience, this directly affects apparent isotropy of light speed. For GPS it was most convenient to choose the ECI "frame" as "rest frame". Consequently clock synchronization makes radio waves appear to propagate isotropically relative to the ECI frame, wrt which the Earth rotates. The relative speed (also called "closing speed") of radio waves and a GPS receiver at rest on the Earth is c-v. This is merely an example that a nominal difference of one-way speeds can be measured; it has no direct bearing on the example of the OP but the same can be done for two one-way light speeds. I gave that example because it was unclear to me what you had in mind and hoped to achieve, but I suspected that it was based on a misunderstanding related to these matters.

Thanks for your response. I appreciate a worldwide navigation system needs a standardized reference frame to work from - ECI in this case. Knowing next to nothing about the actual workings of the GPS network, am thinking when you refer to a relative speed of c-v, this translates as a Doppler shift measurement, inferring relative velocity between that satellite emitter and terrestrial GPS receiver (or extended by computation to some other physical reference object). If I recall right there is normally four such satellites always in view to accurately 'quadrangulate' position and speed. That of itself doesn't allow one-way light speed measurements. As you put in #51, we know SR postulates that (one-way) c is an invariant in any frame, whether light is being emitted within that frame, or received from a source in that or any other frame.
EDIT: Just came across an article "One-Way Light Speed Determination Using the Range Measurement Equation of the GPS"; Stephan J. G. Gift. Looks like I misinterpreted what you have been arguing - sorry. After just a brief skim, I have the impression what the author is claiming is one-way c is actually Sagnac. The consensus view is one-way can't be done:http://math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html#one-way_tests" But then goes on to quote upper bounds!

Because I'm increasingly of the view there is little if any physical sense to a genuine one-way variation in c, might as well give a brief sketch of what was planned, but now canned. For all I know it's old hat but here goes anyway. Basically, two long, closely spaced parallel lengths of say optical fiber are fed by a splitter from a common laser oscillator, and each match terminated at the far end, so there are no reflected waves. The fibres are very slightly different in phase constant, and thus a kind of 'potential interference pattern' exists between adjacent sections of fibers owing to the different guide wavelengths. Connect a sampling probe between the two at or near a node. If there is a one-way c, the location of such a node should be pushed variously towards or away from the laser source depending on relative orientation between fibers and presumably the direction of 'aether flow' (what else could be postulated as a variable c cause?). Moving the arrangement slowly through various angles, while maintaining a null at the probe by means of a variable phase section in one arm of the feed splitter, it should be possible to work out both the magnitude and sense of 'delta c'. Probably not sensitive enough in that form at least, but in principle it should work. No clocks! Most likely just hunting for a ghost, so have no interest in pursuing it further. Is there an obvious fatal design flaw?

 

[harrylin]

Thanks for your response. I appreciate a worldwide navigation system needs a standardized reference frame to work from - ECI in this case. Knowing next to nothing about the actual workings of the GPS network, am thinking when you refer to a relative speed of c-v, this translates as a Doppler shift measurement, inferring relative velocity between that satellite emitter and terrestrial GPS receiver (or extended by computation to some other physical reference object). If I recall right there is normally four such satellites always in view to accurately 'quadrangulate' position and speed. That of itself doesn't allow one-way light speed measurements. As you put in #51, we know SR postulates that (one-way) c is an invariant in any frame, whether light is being emitted within that frame, or received from a source in that or any other frame.
EDIT: Just came across an article "One-Way Light Speed Determination Using the Range Measurement Equation of the GPS"; Stephan J. G. Gift. Looks like I misinterpreted what you have been arguing - sorry. After just a brief skim, I have the impression what the author is claiming is one-way c is actually Sagnac. The consensus view is one-way can't be done:http://math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html#one-way_tests" But then goes on to quote upper bounds!

I found that paper but have reason to suspect that it's not peer reviewed and it's certainly not free of errors - already its main message is wrong. However, the "Sagnac correction" of GPS is effectively the same as the one-way speed of light relative to a moving object (in modern jargon, their "closing speed" c-v).

Because I'm increasingly of the view there is little if any physical sense to a genuine one-way variation in c, might as well give a brief sketch of what was planned, but now canned. For all I know it's old hat but here goes anyway. Basically, two long, closely spaced parallel lengths of say optical fiber are fed by a splitter from a common laser oscillator, and each match terminated at the far end, so there are no reflected waves. The fibres are very slightly different in phase constant, and thus a kind of 'potential interference pattern' exists between adjacent sections of fibers owing to the different guide wavelengths. Connect a sampling probe between the two at or near a node. If there is a one-way c, the location of such a node should be pushed variously towards or away from the laser source depending on relative orientation between fibers and presumably the direction of 'aether flow' (what else could be postulated as a variable c cause?). Moving the arrangement slowly through various angles, while maintaining a null at the probe by means of a variable phase section in one arm of the feed splitter, it should be possible to work out both the magnitude and sense of 'delta c'. Probably not sensitive enough in that form at least, but in principle it should work. No clocks! Most likely just hunting for a ghost, so have no interest in pursuing it further. Is there an obvious fatal design flaw?

- yes it sounds to me a bit like an experiment with cables that was done some years ago in Belgium and from which a positive effect was claimed... let's see the FAQ:

http://www.phys.ncku.edu.tw/mirrors/physicsfaq/Relativity/SR/experiments.html#Experiments_not_consistent_with_SR

["Amateurs look for patterns, professionals look at error bars" - hmm, that's way oversimplified - but that's another topic!]

Ah I found it, something like that was done once by de Witte (regretfully the experiment isn't directly discussed there). Anyway, in theory such experiments will only eventually find effects from the non-inertial motion of the Earth. The Silvertooth experiment there also looked for effects on phases, and was also discredited on theoretical grounds (the alternative theory could not explain the claimed results either). If I correctly recall (for I have now no time to redo the analysis), waves (from one side as well as from two sides) interfere the same when in motion as in rest - perhaps someone else can eventually correct me and elaborate.

 

[Q-reeus]

I found that paper but have reason to suspect that it's not peer reviewed and it's certainly not free of errors - already its main message is wrong. However, the "Sagnac correction" of GPS is effectively the same as the one-way speed of light relative to a moving object (in modern jargon, their "closing speed" c-v).

As things stand you've got me confused, hopefully just a case of unspoken qualifications. In #31 "And yes, truly measuring the one way speed of light is indeed tantamount to measuring absolute velocity.", which I took as saying it was an impossibility. but then in #50 "The relative speed (also called "closing speed") of radio waves and a GPS receiver at rest on the Earth is c-v. This is merely an example that a nominal difference of one-way speeds can be measured..." - Gift's paper was I thought arguing just that. Let's bypass all the technical intricacies of how GPS works and consider physics in some nominally inertial lab frame. A basic consequence of a real nonreciprocal one-way c : from the fundamental relation c = lambda*f, if f is fixed any nonreciprocal c automatically means a nonreciprocal lambda in that frame. This regardless of whether the light source is within that or another frame. Do you agree with this? And contrary to the consensus viewpoint, lambda variation is at least in principle readily measurable (see below).

- yes it sounds to me a bit like an experiment with cables that was done some years ago in Belgium and from which a positive effect was claimed... let's see the FAQ:
http://www.phys.ncku.edu.tw/mirrors/physicsfaq/Relativity/SR/experiments.html#Experiments_not_consistent_with_SR
["Amateurs look for patterns, professionals look at error bars" - hmm, that's way oversimplified - but that's another topic!]

Thanks for that link - more details on one-way tests than the Baez site.

Ah I found it, something like that was done once by de Witte (regretfully the experiment isn't directly discussed there). Anyway, in theory such experiments will only eventually find effects from the non-inertial motion of the Earth.

Not so - see below.

The Silvertooth experiment there also looked for effects on phases, and was also discredited on theoretical grounds (the alternative theory could not explain the claimed results either). If I correctly recall (for I have now no time to redo the analysis), waves (from one side as well as from two sides) interfere the same when in motion as in rest - perhaps someone else can eventually correct me and elaborate.

Found the 1986 paper by Silvertooth - seems OK in principle. Interested though in any actual detailed critique you may know of. A very similar style of differential phase type test was performed by C Navia et al, referenced in the above link. Interesting that the only criticism leveled there was one of sloppy technique - but nothing about fundamental theoretical inadequacy.

Originally Posted by Q-reeus:
grav-universe - my understanding is SR and LET are indistinguishable in any operational sense.

"They are if all frames are completely symmetrical, which so far they appear to be. LET covers a broader scope than SR, however. SR only works if all frames are symmetrical, while LET does not necessarily require it."

That was from #42. I would now have to agree with grav-universe on that point: SR by the 2nd postulate forbids one-way c variation, whereas LET implicitly requires it. My assumption was LT's automatically compensate in both theories. That's only true for two-way measurements and 'normal physics' which depends on invariant two-way c. Hence one-way tests are crucial to distinguish between SR and LET - and differential phase type measurements are perfectly capable in principle of doing that - as per c = lambda*f discussed above. How so? Well my proposed twin fibers setup for one. A nonreciprocal lambda equally stretches/compresses lambda along both lines - and the associated node locations that for sure can be measured; just a question of required sensitivity. Another even simpler example, not very sensitive but 'for sure' in principle - a single match terminated line containing one or more sections of a very low Q almost '1/4 lambda' transmission cavity. Won't go into the details here, but not hard to show that any non-reciprocity in c and thus lambda will result in a corresponding reflection phase and amplitude variation. Once we see past the false dilemma of 'clock sync' and realize phase methods completely bypass this issue, 'impossibility of one-way c measurement' now looks pretty stupid imho. I've done a 180 degree flip on outlook here. Desperately resisting being sucked into a SR/LET theoretic dispute BH!

 

[ghwellsjr]

I would now have to agree with grav-universe on that point: SR by the 2nd postulate forbids one-way c variation, whereas LET implicitly requires it. My assumption was LT's automatically compensate in both theories. That's only true for two-way measurements and 'normal physics' which depends on invariant two-way c. Hence one-way tests are crucial to distinguish between SR and LET - and differential phase type measurements are perfectly capable in principle of doing that - as per c = lambda*f discussed above. How so? Well my proposed twin fibers setup for one. A nonreciprocal lambda equally stretches/compresses lambda along both lines - and the associated node locations that for sure can be measured; just a question of required sensitivity. Another even simpler example, not very sensitive but 'for sure' in principle - a single match terminated line containing one or more sections of a very low Q almost '1/4 lambda' transmission cavity. Won't go into the details here, but not hard to show that any non-reciprocity in c and thus lambda will result in a corresponding reflection phase and amplitude variation. Once we see past the false dilemma of 'clock sync' and realize phase methods completely bypass this issue, 'impossibility of one-way c measurement' now looks pretty stupid imho. I've done a 180 degree flip on outlook here. Desperately resisting being sucked into a SR/LET theoretic dispute BH!

The only difference between LET and SR is a philosophical one; LET claims that nature operates on a single unidentifiable absolute ether rest frame in which we are almost always moving with respect to, while SR claims that every inertial observer can consider himself to be at rest in, what amounts to, the absolute ether rest frame. There can be no test that can distinguish between the two.

 

[Q-reeus]

The only difference between LET and SR is a philosophical one; LET claims that nature operates on a single unidentifiable absolute ether rest frame in which we are almost always moving with respect to, while SR claims that every inertial observer can consider himself to be at rest in, what amounts to, the absolute ether rest frame. There can be no test that can distinguish between the two.

That's what I've thought till now. As for LET reference frame, 'absolute' need not mean that as such - I'm rather partial to an increasing view that CMBR provides a reference gauge for a 'locally absolute' rest frame. Think of the analogous 2D balloon model - any point locally stationary on the skin can serve as a local rest frame, though no globally 'absolute rest frame' is possible. But really I think it gets down to whether non-reciprocal c is in principle detectable - currently I would say yes - via properly designed one-way differential phase measurements. If you know of a reference that can shoot that idea down, please post it!

 

[harrylin]

As things stand you've got me confused, hopefully just a case of unspoken qualifications. In #31 "And yes, truly measuring the one way speed of light is indeed tantamount to measuring absolute velocity.", which I took as saying it was an impossibility. but then in #50 "The relative speed (also called "closing speed") of radio waves and a GPS receiver at rest on the Earth is c-v. This is merely an example that a nominal difference of one-way speeds can be measured..." - Gift's paper was I thought arguing just that. Let's bypass all the technical intricacies of how GPS works and consider physics in some nominally inertial lab frame. A basic consequence of a real nonreciprocal one-way c : from the fundamental relation c = lambda*f, if f is fixed any nonreciprocal c automatically means a nonreciprocal lambda in that frame. This regardless of whether the light source is within that or another frame. Do you agree with this? And contrary to the consensus viewpoint, lambda variation is at least in principle readily measurable (see below).

Thanks for that link - more details on one-way tests than the Baez site.

Not so - see below.

Found the 1986 paper by Silvertooth - seems OK in principle. Interested though in any actual detailed critique you may know of. A very similar style of differential phase type test was performed by C Navia et al, referenced in the above link. Interesting that the only criticism leveled there was one of sloppy technique - but nothing about fundamental theoretical inadequacy.

Aargh - I spent one hour on replying (basically it was all "no" + long explanations) but lost it all because I didn't make a backup and this site threw me out... too bad, won't have time anymore and should be sleeping now. If you still have any question one or two days from now, please ask again!

 

[ZapperZ]

Er.. not sure if anyone brought up this link already (I didn't read through all 4 pages of this discussion), but in case it hasn't, one might want to look at this review:

http://arxiv.org/abs/1011.1318

Zz.

 

[Q-reeus]

Er.. not sure if anyone brought up this link already (I didn't read through all 4 pages of this discussion), but in case it hasn't, one might want to look at this review:

http://arxiv.org/abs/1011.1318

Zz.

Yes it was listed early on in the thread, but I never got to read all of it myself. They claim clock synch is no barrier to making one-way tests, and quote Clifford Will to that effect. On the other hand these reviewers deny it can truly be done; ie. 'one-way' is always de facto 'two-way': http://math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html#one-way_tests"

 

[Q-reeus]

Aargh - I spent one hour on replying (basically it was all "no" + long explanations) but lost it all because I didn't make a backup and this site threw me out... too bad, won't have time anymore and should be sleeping now. If you still have any question one or two days from now, please ask again!

Amazing coincidence - precisely same thing happened to me (Microsoft automatic update = unwelcome reboot!) Ditto way overdue for sleep here too.

EDIT: In #55 "Another even simpler example, not very sensitive but 'for sure' in principle - a single match terminated line containing one or more sections of a very low Q almost '1/4 lambda' transmission cavity. Won't go into the details here, but not hard to show that any non-reciprocity in c and thus lambda will result in a corresponding reflection phase and amplitude variation." That was just outright wrong and is withdrawn unreservedly. Reflection always implies a two-way measurement!

 

[harrylin]

Yes it was listed early on in the thread, but I never got to read all of it myself. They claim clock synch is no barrier to making one-way tests, and quote Clifford Will to that effect. On the other hand these reviewers deny it can truly be done; ie. 'one-way' is always de facto 'two-way': http://math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html#one-way_tests"

Both are somewhat correct: if a one-way experiment can break the PoR, then in principle a suitably designed two-way experiment should be capable of the same. However, different set-ups are more suited to test other aspects of the theory. In particular the repeat of the Marinov one-way experiment will be interesting.

 

[harrylin]

Amazing coincidence - precisely same thing happened to me (Microsoft automatic update = unwelcome reboot!) Ditto way overdue for sleep here too.

EDIT: In #55 "Another even simpler example, not very sensitive but 'for sure' in principle - a single match terminated line containing one or more sections of a very low Q almost '1/4 lambda' transmission cavity. Won't go into the details here, but not hard to show that any non-reciprocity in c and thus lambda will result in a corresponding reflection phase and amplitude variation." That was just outright wrong and is withdrawn unreservedly. Reflection always implies a two-way measurement!

OK. Just a short precision about my unclear explanation: with "nominal c-v" I meant that that measurement is merely a result of the observer's reference and synchronization -> you can transform it away to v=0 in which case you obtain isotropic light speed (see dictionary.com, "nominal" ).

Cheers,
Harald

 

[Q-reeus]

OK. Just a short precision about my unclear explanation: with "nominal c-v" I meant that that measurement is merely a result of the observer's reference and synchronization -> you can transform it away to v=0 in which case you obtain isotropic light speed (see dictionary.com, "nominal" ).
Cheers, Harald

Just knew it had to be something like that; glad we see eye to eye on that one! Meanwhile I've had an epiphany event of sorts.
ghwellsjr wrote in #56:
"The only difference between LET and SR is a philosophical one; LET claims that nature operates on a single unidentifiable absolute ether rest frame in which we are almost always moving with respect to, while SR claims that every inertial observer can consider himself to be at rest in, what amounts to, the absolute ether rest frame. There can be no test that can distinguish between the two."
Yep, now agree.
harrylin wrote in #62:
"Both are somewhat correct: if a one-way experiment can break the PoR, then in principle a suitably designed two-way experiment should be capable of the same. However, different set-ups are more suited to test other aspects of the theory. In particular the repeat of the Marinov one-way experiment will be interesting."
Yes and yes and yes.

Focusing on means to measure a notional one-way c, hadn't stop to think what a non-null result would fully imply. In the proposed twin fiber arrangement of #53, if null balancing was foregone, as the interference pattern shifted with orientation, the coupling probe will act as a variable scatterer/reflector - changing the overall energy flow and thus the physics. Consequently we can draw the conclusion a finite one-way c effect is automatically incompatible with the basic postulate of SR/LET - physical equivalence of all inertial reference frames. Way too much observational support for that to be in question, at anything above Planck scale physics anyway. So another 180 degree turn and it's back full circle. No point in hunting for a non-event. I'm exhausted - catch you all much later!