Clock synchronization and relativity of simultaneity

[Fantasist]

I addressed already recently in this thread the issue of defining the synchronicity of clocks moving relatively to each other (considering that the synchronization by Einstein's method implies clocks at rest), but it occurred to me now that even for clocks at rest relatively to each other there is a problem with the practical definition as far as its frame dependence is concerned. On the basis of his two-way light signal propagation thought experiment, Einstein concludes at the end of paragraph 2 in http://www.fourmilab.ch/etexts/einstein/specrel/www/

So we see that we cannot attach any absolute signification to the concept of simultaneity, but that two events which, viewed from a system of co-ordinates, are simultaneous, can no longer be looked upon as simultaneous events when envisaged from a system which is in motion relatively to that system.

Let us look a little bit closer at the thought experiment on which this conclusion is based. For this purpose, let us slightly modify it for clarity and assume each of the two clocks (stationary with regard to each other) sends out a light signal at the same time (according to each of the clocks). Halfway between the two clocks is a detector that registers both signals. We can now define that the two clocks in question are synchronized with each other if the detector registers them simultaneously (according to its own time). Let's further assume that in this case both signals are completely absorbed by the detector. On the other hand, if the signals do not arrive simultaneously (within a defined window) they are not absorbed but carry on to the other clock (where they can be subsequently detected).
Now with this practical definition of simultaneity, how can this possibly be frame dependent? The two signals are either absorbed or not absorbed. All observers would have to agree about this physical fact. So evidently, this setup could not experimentally define the relativity of simultaneity. The question is how do we have to change/generalize the setup so that it is consistent with Einstein's conclusion?

 

[A.T.]

We can now define that the two clocks in question are synchronized with each other if the detector registers them simultaneously (according to its own time).

According to any frame. The detection of the signals is one event. the way you set it up. Relativity of simultaneity requires spatial separation of the events.

 

[Dale]

Now with this practical definition of simultaneity, how can this possibly be frame dependent?

The length of the optical path to and from the detector is frame dependent. The simultaneous arrival and unblocking will be proof that the signals were not emitted simultaneously in other frames.

 

[Nugatory]

Now with this practical definition of simultaneity, how can this possibly be frame dependent?

All observers in all frames will agree about whether the two light signals met at the central detector or didn't meet at the central detector.

What is frame-dependent here is whether the two light signals were emitted at the same time. If the two light signals do meet at the central detector an observer who is at rest relative to the central detector will find that the emission events were simultaneous, but observers moving relative to it will find that they were emitted at different times.

 

[Fantasist]

According to any frame. The detection of the signals is one event. the way you set it up. Relativity of simultaneity requires spatial separation of the events.

The events here are the ticks of each of the clocks, so they are spatially separated. It is only that the according timestamps are analyzed in one place as to determine whether they are actually simultaneous or not.

Note that this setup is essentially the same as Einstein's, only that the light signal is not reflected but the second clock emits its own signal and the two are then timed at a single third location (thus avoiding the problem of getting the time at the reflector back to the originating clock).

 

[Fantasist]

All observers in all frames will agree about whether the two light signals met at the central detector or didn't meet at the central detector.

What is frame-dependent here is whether the two light signals were emitted at the same time. If the two light signals do meet at the central detector an observer who is at rest relative to the central detector will find that the emission events were simultaneous, but observers moving relative to it will find that they were emitted at different times.

OK, but it was our definition that the original events are simultaneous if the signals meet at the central detector. How would you change this to make this observer dependent?

 

[Nugatory]

OK, but it was our definition that the original events are simultaneous if the signals meet at the central detector. How would you change this to make this observer dependent?

(You say "observer-dependent", although "frame-dependent" would be more accurate)

We don't need to. You've specified that the two signals are emitted simultaneously according to two clocks that are at rest with one another and synchronized in the frame in which they are at rest. This condition is equivalent to saying that the signals will meet at a centrally located detector also at rest relative to the clocks; so we have the frame-independent but uninteresting tautology that if two events are simultaneous in a given frame then they are simultaneous in that frame.

 

[A.T.]

The events here are the ticks of each of the clocks, so they are spatially separated.

The tick events aren't synchronized in other frames. Only the detection events, because they a identical or just one event.

 

[Dale]

OK, but it was our definition that the original events are simultaneous if the signals meet at the central detector. How would you change this to make this observer dependent?

It is not the standard definition. Furthermore, it is a bad definition because the speed of light becomes frame dependent and anisotropic.

See my previous reply. The optical path lengths are frame variant. So if you fix the time by definition then you get a variable speed of light.

 

[Fantasist]

We don't need to. You've specified that the two signals are emitted simultaneously according to two clocks that are at rest with one another and synchronized in the frame in which they are at rest.

I did not say the two signals are emitted simultaneously but only 'at the same time' i.e when each of the clocks shows a given reading. The corresponding events are only simultaneous if the clocks are synchronized. And you need some procedure to define simultaneity/synchronization, like the one described by Einstein (comparing the timings of a reflected light signal) or my version with comparing the reception time of the two light signals by a detector halfway between the two clocks (you could make up any number of other ways of doing it).

This condition is equivalent to saying that the signals will meet at a centrally located detector also at rest relative to the clocks; so we have the frame-independent but uninteresting tautology that if two events are simultaneous in a given frame then they are simultaneous in that frame.

It is not a tautology. The conclusion is (given that we have defined simultaneity as suggested above)

The light signals meet at the central detector <=> The clocks are synchronized

Now you said yourself that in this case the light signals meet at the central detector in any reference frame, but if the clocks are not synchronized in any reference frame, the above conclusion would become obviously false. Which means that we require a different procedure by means of which a moving observer can decide whether the clocks are synchronized or not.

 

[Fantasist]

It is not the standard definition. Furthermore, it is a bad definition because the speed of light becomes frame dependent and anisotropic.

See my previous reply. The optical path lengths are frame variant. So if you fix the time by definition then you get a variable speed of light.

The question is how do you practically define the speed of light for a moving observer/reference frame? In the rest frame of the two clocks it is easy: if the clocks are synchronized and you know their distance, you can get the speed of light from the recorded timestamps of the emission and reception. But how would the timestamps be recorded in a different reference frame?

 

[A.T.]

But how would the timestamps be recorded in a different reference frame?

The other reference frame (where your emitters are moving) has its own resting synchronized clocks, which record the timings of the emission/reception events.

 

[Dale]

The question is how do you practically define the speed of light for a moving observer/reference frame? In the rest frame of the two clocks it is easy

In any inertial frame it is easy. The speed of light is c.

I think what you really want to know is how can you define a simultaneity convention which is compatible with the speed of light being c. Einstein already did that.

 

[Fantasist]

The other reference frame (where your emitters are moving) has its own resting synchronized clocks, which record the timings of the emission/reception events.

The only problem is that these events have nothing to do with the events we are interested in (which are related to the clocks/detectors defining the original reference frame).

 

[A.T.]

The only problem is that these events have nothing to do with the events we are interested in (which are related to the clocks/detectors defining the original reference frame).

They are the same physical events. Different frames merely assign different space-time coordinates to them.

 

[Dale]

@Fantasist Let me see if I can restate what I think you are trying to do.

Given a reference frame O and two clocks a and b both moving at speed v in O, you want to find some simple experiment to determine if those clocks are synchronized in O, not in their rest frame.

 

[Fantasist]

They are the same physical events. Different frames merely assign different space-time coordinates to them.

The events we are talking about here are tied into the clocks/detectors defining the reference frame (namely when these register the emitted light signal), and as you pointed out above yourself, these are physically different sets of clocks. So how could the associated detection events be physically the same?

 

[A.T.]

The events we are talking about here are tied into the clocks/detectors defining the reference frame (namely when these register the emitted light signal), and as you pointed out above yourself, these are physically different sets of clocks. So how could the associated detection events be physically the same?

Only the clocks resting in the original frame emit signals. The clocks resting in the other frame merely time stamp these emission events for the other frame.

Is DaleSpam's suggestion in post #16 what you are trying to ask?

 

[Fantasist]

@Fantasist Let me see if I can restate what I think you are trying to do.

Given a reference frame O and two clocks a and b both moving at speed v in O, you want to find some simple experiment to determine if those clocks are synchronized in O, not in their rest frame.

Essentially, yes. If we have an experimental procedure that defines the synchronization of the clocks a and b in their rest frame, then we must also have an experimental procedure that defines the synchronization of the same clocks a and b in O. Otherwise we would not be entitled to make any statement about their synchronization in the latter frame.

 

[Fantasist]

Only the clocks resting in the original frame emit signals. The clocks resting in the other frame merely time stamp these emission events for the other frame.

That would mean you couldn't determine the speed of light in one frame without knowing the events defined by the clocks of the other frame.

 

[A.T.]

If we have an experimental procedure that defines the synchronization of the clocks a and b in their rest frame, then we must also have an experimental procedure that defines the synchronization of the same clocks a and b in O.

You know their distance in O, so just place two clock resetters at rest in O that distance apart.

 

[Dale]

Essentially, yes. If we have an experimental procedure that defines the synchronization of the clocks a and b in their rest frame, then we must also have an experimental procedure that defines the synchronization of the same clocks a and b in O.

Sure. A new experimental procedure is not needed. Simply determine the "the time of the stationary system" O, as defined by Einstein, and then a and b are synchronized in O if they read the same value at the same time in O.

Remember, this is how we concluded the previous discussion.

 

[Fantasist]

Sure. A new experimental procedure is not needed. Simply determine the "the time of the stationary system" O, as defined by Einstein, and then a and b are synchronized in O if they read the same value at the same time in O.

Remember, this is how we concluded the previous discussion.

Well, obviously, if the clocks read the same value they synchronize; the question is do they read the same value or not (given the assumptions)? Since the discussion seems to have lost focus of this question, let me address the problem again using Einstein's own definitions:

in paragraph 1 of his 1905 paper, he defines the synchronization condition in the rest frame of the clocks A and B with the help of the emission, reflection and return timings of a light signal in the form

(1) t_A2 - t_B = t_B - t_A1

( where I have used t_A1 and t_A2 instead of t_A and t_A' )

As one can imagine these timings being logged at each clock, they are therefore unambiguous and frame independent (as was agreed by everybody here).

Now in paragraph 2, Einstein considers the situation from a frame O moving with velocity v relatively to the clocks A and B, and claims that in this reference frame the timings would be

(2) t_B - t_A1 = r_AB/(c-v) ; t_A2 - t_B = r_AB/(c+v)

(where r_AB is the distance of the clocks A and B in the frame where they appear to be moving).

Now as written, Eqs (1) and (2) are obviously algebraically inconsistent (unless v=0), so the timing variables in (2) can actually not be the same as those in (1). Strictly speaking he should have therefore written for instance

(2a) t_B' - t_A1' = r_AB/(c-v) ; t_A2' - t_B' = r_AB/(c+v)

where the prime indicates that timings are taken in frame O.

The question is where are these primed timings taken and logged? It can obviously not be the clocks A and B themselves as we have agreed they record the unprimed timings. But if they are recorded elsewhere, we have no right to associate these timings with the clocks A and B anymore. So one way or another one appears to run into contradictions here.

 

[Dale]

so the timing variables in (2) can actually not be the same as those in (1). Strictly speaking he should have therefore written for instance

So your complaint is about his notation? OK. Yes, in (1) the times refer to the proper time on clocks A and B whereas in (2) the times refer to the coordinate time in the stationary system at clocks A and B.

Since he clarifies what each means in the text this does not seem to be a substantive complaint.

The question is where are these primed timings taken and logged?

He very clearly describes that in the text. The clocks A and B in section 2 display "time of the stationary system".

They are synchronous in the stationary system by definition, but the experiment of section 1 shows that they are not synchronous in the moving system.

 

[Fantasist]

He very clearly describes that in the text. The clocks A and B in section 2 display "time of the stationary system".
They are synchronous in the stationary system by definition, but the experiment of section 1 shows that they are not synchronous in the moving system.

Sorry, but I don't see how you can get to this conclusion. The timings in the stationary system are based on stationary clocks/detectors and how the path of the moving clock and the light signal is registered by these. The synchronization of the moving clocks in their own system does not come into it at all. They could be synchronized or de-synchronized as you want, the timings in the stationary system are not affected by this (as you agreed above, the corresponding timing variables should actually be different variables to those representing the timings in the moving frame).

 

[PeterDonis]

The synchronization of the moving clocks in their own system does not come into it at all.

They most certainly do if you are trying to figure out what is "synchronous in the moving system", which is what DaleSpam said. "Synchronous in the moving system" is defined by the synchronization of the moving clocks in their own system.

 

[John F. Gogo]

It is not the standard definition. Furthermore, it is a bad definition because the speed of light becomes frame dependent and anisotropic.

See my previous reply. The optical path lengths are frame variant. So if you fix the time by definition then you get a variable speed of light.

Yes, the two-way, one-clock, single location measure is convention. One-way anisotropic measure would require the unified coordination between the sender and receiver.

 

[Dale]

The synchronization of the moving clocks in their own system does not come into it at all. They could be synchronized or de-synchronized as you want

This is incorrect. Since the moving clocks are (by design) synchronized in the stationary system they cannot be synchronized in the moving frame, regardless of what you might want.

 

[ash64449]

If the clocks which are separated by a certain distance are synchronized in their rest frame,then they are not synchronized in a frame where they are seen to be moving... In fact the clock at the rear is ahead of the clock at the front by the amount $${ \frac{L_0v}{c^2}}$$ where $$L_0$$ is the rest separation between the clocks and $$v$$ the velocity of the clocks.

Edit: why are my LaTex at the centre?

 

[ash64449]

shouldn't the clock synchronization be carried in their own rest frame? i mean observers who are there in that frame itself.

They do this to understand the timings of the different events in their own frame... the clock nearer to the event measures this and then it becomes the time at which that event occurred in that frame..

 

[Ibix]

If I'm understanding right, fantasist, you're defining two clocks to be synchronised if identically time-stamped pulses arrive simultaneously at the point half-way between the clocks.

Where relativity of simultaneity comes from is the realisation that "half-way between" means different things in different frames. This is the point of Einstein's train struck by lightning at both ends. According to an observer on the platform, "half-way between" the strikes is the middle of the platform; to an observer on the train it is the middle of the train. These are not the same place. And since you cannot have the light from the strikes arriving simultaneously at the center of the train and also simultaneously at the center of the platform, one or other must say that the strikes were not simultaneous.

Everyone agrees that the light from the strikes arrives simultaneously at the center of the platform. But since they don't agree that the center of the platform is half-way between the strikes, they don't think that is relevant to their definition of simultaneity.

The same is true of your clocks. Everyone agrees that the pulses arrive simultaneously at your detector. But since they don't agree that the detector is at the point half-way between the emission points, this is not relevant to them.

 

[DrGreg]

Edit: why are my LaTex at the centre?

Use "itex" instead of "tex" if you want inline LaTeX.

 

[Dale]

shouldn't the clock synchronization be carried in their own rest frame? i mean observers who are there in that frame itself.

They do this to understand the timings of the different events in their own frame... the clock nearer to the event measures this and then it becomes the time at which that event occurred in that frame..

Yes, this is how it is done in Einsteins paper.

He then shows how doing so leads to something that does not meet the definition of synchronization in the moving frame.

 

[Fantasist]

If I'm understanding right, fantasist, you're defining two clocks to be synchronised if identically time-stamped pulses arrive simultaneously at the point half-way between the clocks.
Where relativity of simultaneity comes from is the realiation that "half-way between" means different things in different frames. .

If the detector half-way between the clocks is co-moving with latter, then it is half-way between them in any reference frame.

Anyway, to avoid confusion, note that we have been switching here back and forth between Einstein's original version of the synchronization procedure (which doesn't involve a separate detector but instead a signal round-trip) and my own version. The last few posts have actually been dealing more with the former rather than the latter.

 

[Fantasist]

If the clocks which are separated by a certain distance are synchronized in their rest frame,then they are not synchronized in a frame where they are seen to be moving... In fact the clock at the rear is ahead of the clock at the front by the amount $${ \frac{L_0v}{c^2}}$$ where $$L_0$$ is the rest separation between the clocks and $$v$$ the velocity of the clocks.

Note that we should not have to resort to the Lorentz transformation formula to prove the frame dependence of the clock synchronization. Einstein claims it already when he defines his clock synchronization formula in paragraphs 2&3 in his paper, before he even starts to derive the Lorentz transformation. It is something that should follow from his definitions alone.