# The synchronization procedure

1. Feb 15, 2008

### ehj

The idea is that you have a clock at the origin of some frame O. In order to synchronize this clock with 2 other clocks placed at x = -X and x = X (one dimensional) these two clocks are fixed to show the time t = x/c and they start ticking when they recieve a light flash from when the clock at the origin when it is started.
My question then is, doesn't this neglect the fact that an observer at the origin O needs to wait for the information about the "starting" of the clocks at x = -X and x = X? Won't these two clocks be delayed with t = x/c ? I would think you would need to fix the two clocks for times given by t = 2x/c?
I'm probably wrong, so please explain in what way the clocks are synchronized.

2. Feb 15, 2008

### yuiop

Yes, you have to allow for light propagation delays.

Say we send a radar signal from the origin to the +x clock and the return signals takes 2 seconds for the round trip then we assume the distance x is 1 light second. We pause both clocks and set the +x clock to +1 seconds and the clock at the origin to 0 seconds. We then start the origin clock and simultaneously send a start signal to the +x clock which is preset at 1 seconds. When the start signal arrives at the +x clock and starts it ticking, both the +x clock and the clock at the origin read 1 second and are syncronised.

3. Feb 16, 2008

### ehj

So when assigning a coordinate to an event, you must take the delay into consideration?
Let's say that an observer at the origin of a frame observes two clocks, one at the distance of X and another at the distance 2X for instance. If the observer (stationary to the clocks) observes the watches as being synchronized, they in reality aren't because the signal he sees is delayed, and therefore wont have same time coordinates?

4. Feb 16, 2008

### K.J.Healey

No, in reality they ARE (and by reality I mean HIS reference frame).

There is no REALITY reference frame, you are required to choose one. In his reference frame they are synched, but in both clocks ref frames they are not. Who is right? Is that even a valid question to ask? I don't think so.

You are required to choose a reference frame, and then only talk about observables in that frame.

5. Feb 16, 2008

### Staff: Mentor

Absolutely. Assuming, of course, that the "event" you would like to assign coordinates to is the emission of the light. (Reception of the light can also be an event.)

Right. In order to interpret the raw "observations" you must always take the light travel time into account.

6. Feb 16, 2008

### ehj

Thanks doc, you answered my question!

7. Feb 16, 2008

### bernhard.rothenstein

synchronization

Is there something wrong when I consider a ticking clock C(0) located at the origin O of I and a stopped clock C(x) located at the point M(x) and fixed to diplay a t=x/c time. A light signal starts from O towsrds clock C(x) starting it. I think they are synchronized a la Einstein.

8. Feb 17, 2008

### Staff: Mentor

There's nothing wrong with setting the clocks fixed at t = x/c and triggering them to start when a light signal from the clock at the origin (emitted at t = 0) reaches them. They will be perfectly synchronized in the usual manner. (Don't set them to to t = 2x/c.) The observer at the origin doesn't have to wait for any information from the clocks--it's all arranged in advance; light travel time has already been accounted for. (If you set them at t = 2x/c, then the clocks would merely appear synchronized with the clock at the origin from the view of the observer at the origin--but he knows better! He knows that to interpret what he sees, he must take into account light travel time.)

I agree.