Compelling evidence for in vacuo light speed anisotrophy
in [Relativity]
|
From: Edward Green on 19 Apr 2008 09:24 On Apr 13, 8:47 pm, Yuan...(a)gmail.com wrote: > On Apr 13, 12:17 pm, Edward Green <spamspamsp...(a)netzero.com> wrote: > > > On Apr 13, 12:03 pm, Yuan...(a)gmail.com wrote: > > > > Actually, he asked a question. He wasn't making a logical proposition. > > > I was aware of that when I wrote, and also that somebody would > > probably make an issue of it. Congratulations. :-) > > Well it seems to me that it's issues that we should be discussing. > > If not issues, then what? > > Non issues? This _is_ a non-issue. > > There again, this is physics, not logic. The two are only feebly > > > connected. > > > Quite strongly, I would say -- unless you mean this as a sociological > > observation/insult against physicists. > > Not at all! > > Logic is about reasoning. Physics is far more than that. It's about reasoning about observation. That's why I said "quite strongly" connected, and not "identical". <...> > Apparently you didn't understand well enough, or are not courageous > enough, to answer the question. Nothing to start the day like a nice insult! Same to you. > Another question for you to run from. Failing to play boring rhetorical games in not "running". > "If the speed of light is isotropic in all frames, then any > observation in conflict with that proposition must be flawed." > > Do you agree with that statement or not? > > In my opinion, it's good logic, but atrocious physics. If I said I agreed you'd probably snip the first part of the trope: If A is true, than observations in conflict with A must be false. It's perfectly good "physics", so long as we keep the whole thing. As a whole, it says nothing about whether A is true or false.
From: Tom Roberts on 19 Apr 2008 13:10 wbbrdr(a)gmail.com wrote: > Error analysis that is based on a false premise can also lead to good > data being painted as bad. But not an error analysis without any such "false premise". Indeed, my error analysis of Miller's algorithm is without any "premise" whatsoever, except that mathematics applies. It is just basic statistics applied to the average of multiple data points, and the errorbars do not depend on whether the variations were due to measurement errors or due to "signal fluctuations" -- the orientation dependence is not significantly different from zero, and it is that INSIGNIFICANT orientation dependence that Miller used to obtain his "absolute motion". > I have explained how that happened with Tom Robert's analysis of > Miller's data. No, you have expressed a fantasy that is devoid of any rationale. If your fantasy were valid, then you would be forced to reject ALL of Miller's conclusions, and Joos's and those of EVERY OTHER experiment. Your fantasy is not science -- it cannot possibly be falsified by experiment (no matter what result is given you can claim "the signal fluctuated that way"; note this applies to the data you like as well as to the data you don't). > [about rejecting Joos data] Just "telling" what you (he) did does not excuse ignoring 21 out of 22 runs. Nobody in their right mind would accept such an "analysis" -- with such a high rate of "failure" the entire technique is called into question. Joos ignored 1 out of 22 runs, which is FAR more believable, especially since he gave a reason OTHER than "it doesn't fit my pre-conceived notion of what the data 'ought to" look like". What Cahill did here is not science. Tom Roberts
From: Tom Roberts on 19 Apr 2008 13:22 Surfer wrote: > The location of Tom Robert's false premise: > The caption under Fig 3. says: > "The assumed-linear systematic drift from the data of Fig. 1. > The lines are between successive Marker 1 values and the points are > Marker 9. These markers are 180 degrees apart, so any real signal has > the same value for every corner and every point; the variations are > purely an instrumentation effect." > > This statement is FALSE, because measurements at Marker 1 and Marker 9 > were not made simultaneously. So any real FLUCTUATING signal would > have different values at the two markers. This is MILLER'S MODEL, in which there is a definite and constant signal. This figure shows how inadequate is his approach of assuming a linear systematic drift. If the signal were "fluctuating" significantly during each turn, then Miller's approach is inadequate, and the entire experiment must be rejected, along with all other such experiments. If it were truly "fluctuating", and not systematically drifting, then Fig 4 would not have the SYSTEMATIC structure it has. And Fig 9 would not have the incredibly-similar plots for each orientation. And Fig 2 would not have such a clear large-scale drift. These data are CLEARLY dominated by a SYSTEMATIC DRIFT and not by "fluctuations". >> what I did is to apply standard >> statistical techniques to the average of a set of data points to derive >> an errorbar on the average. >> > How could you get an valid errorbar doing that, if every data point > just happened to be dead accurate? The errorbar is STILL VALID, as far as computing the orientation dependence of the AVERAGES is concerned. No matter what caused the variations in the data points, the AVERAGES have the large errorbars shown in Fig 5 -- there is no way to escape them. That means that the ORIENTATION DEPENDENCE OF THE AVERAGES is not significantly different from zero, and it is that ORIENTATION DEPENDENCE OF THE AVERAGES which Miller used to obtain his "absolute motion" -- so his "absolute motion" is not significantly different from zero. Hence Cahill's use of Miller's result is completely unwarranted. > You must have made some assumption as to how to distinguish signal > from error. No! No such "assumption is needed". All one needs to do is note that Miller averaged his data, and apply the mathematical consequences of such averaging. The errorbars are undeniable. You live in a fantasy world, and need to learn about basic experimental technique. Tom Roberts
From: Tom Roberts on 19 Apr 2008 13:38 Florian wrote: > I think there is a way to discriminate between a noisy regular signal > that is accurately measured and pure noise. Only if you already know what the signal looks like. This is the basis for synchronous detection and similar techniques. In Miller's model, the shape of the signal is known: C1*sin((2*pi/8)*i+C2), where i is the marker number modulo 8 (i.e. marker 9 = marker 1, etc.), and C1 and C2 are unknown constants for each run. > You have to check the variation of amplitude of the errorbars. If the > amplitude remains constant, then there is a good chance that there is a > noisy signal. It's more complicated than that. With prior knowledge of the signal shape, one can project out the signal and fit the above formula to the data, thus determining C1 and C2. Equivalently, for a signal of the above form, one can perform a digital Fourier transform on the data and look at the amplitude in the signal bin. That is Fig 6 in my paper, and indeed the 1/2-turn bin (# 40) has a small excess compared to neighboring bins. I'll discuss that in the thread with subject "Surprisingly, Tom Robert's paper vindicates Miller". > For example, looking at Fig 5 (*), I would say that there is a noisy > signal. But you have to calculate the variation of the errorbars to > verify it. You can look at the variation of the errorbars in Fig 5. The errorbars on the individual points were of course computed independently. They are remarkably similar in value -- of course this is expected if the interferometer drifts in time independent of orientation, as the model of section IV shows. Tom Roberts
From: Surfer on 20 Apr 2008 00:06
On Apr 20, 2:10 am, Tom Roberts <tjroberts...(a)sbcglobal.net> wrote: > wbb...(a)gmail.com wrote: > > Error analysis that is based on a false premise can also lead to good > > data being painted as bad. > > But not an error analysis without any such "false premise". Indeed, my > error analysis of Miller's algorithm is without any "premise" > whatsoever, except that mathematics applies. > Maths applied without any premises about physical reality, could be divorced from reality. However the caption under Figure 3 shows that you did make a premise. ========================== In: http://www.arxiv.org/abs/physics/0608238 The caption under Figure 3 says: "The assumed-linear systematic drift from the data of Fig. 1. The lines are between successive Marker 1 values and the points are Marker 9. These markers are 180 degrees apart, so any real signal has the same value for every corner and every point the variations are purely an instrumentation effect." This statement is FALSE, because measurements at Marker 1 and Marker 9 were not made simultaneously. So any real FLUCTUATING signal would have different values at the two markers. Light-speed anisotropy caused by motion through a calm and stable medium would give rise to a non-fluctuating signal, but while its reasonable to suppose that light does propagate through a medium, there is NO guarantee that such a medium would be calm and stable. Hence if we wish to discover if a medium exists, the assumption that it is calm and stable would confound our analysis if in reality it were not calm and stable. ========================= > > It is just basic statistics > applied to the average of multiple data points, and the errorbars do not > depend on whether the variations were due to measurement errors or due > to "signal fluctuations" > I find that very hard to believe. > > -- the orientation dependence is not > significantly different from zero, and it is that INSIGNIFICANT > orientation dependence that Miller used to obtain his "absolute motion". > Fig 6 of your paper shows a DFT of raw data. If the signal that Miller was measuring had no orientation dependence, or did not exist, then there should be nothing special about bin 40, which has a period of 1/2 turn. However if a signal existed that DID have an orientation dependence, then bin 40 should stand out. The figure clearly shows that bin 40 STANDS OUT. Ergo, Miller measured a signal WITH an orientation dependence--and it was a fluctuating one. > > > I have explained how that happened with Tom Robert's analysis of > > Miller's data. > > No, you have expressed a fantasy that is devoid of any rationale. If > your fantasy were valid, then you would be forced to reject ALL of > Miller's conclusions, and Joos's and those of EVERY OTHER experiment. > > Your fantasy is not science > >-- it cannot possibly be falsified by > experiment (no matter what result is given you can claim "the signal > fluctuated that way"; note this applies to the data you like as well as > to the data you don't). > Many phenomena that fluctuate in an unpredicatable manner, exhibit regularities of behaviour if they are observed over a period of time. For example gusts of wind are unpredictable, but that has not stopped science developing models of weather. Such models can be falsified by observations. I don't see why the situation should not be the same with light-speed anisotropy. If the anisotropy fluctuates then we should make detailed measurements to characterise the fluctuations and then build a model to account for them. > > > [about rejecting Joos data] > > Just "telling" what you (he) did does not excuse ignoring 21 out of 22 > runs. > You claim that Cahill "ignored" 21 out of 22 runs but his paper clearly acknowledges the 21 runs you refer to. Since he acknowledged them, he didn't ignore them. > > Nobody in their right mind would accept such an "analysis" -- with > such a high rate of "failure" the entire technique is called into > question. > Light speed anisotropy cannot be directly detected in vacuum Joos used helium in his interferometer rather than vacuum. But because of the low refractive index of helium, the optical conditions were close to vacuum. So unsurprisingly, there was a high rate of failure. |