I mentioned in my previous post that “I submitted a paper with a more thorough explanation [of the apparent super-luminal neutrinos detected in the CERN-OPERA experiment] to ArXiv a week ago”. As I write, ArXiv are refusing to publish the paper, without giving reasons in sufficient detail for me to take any corrective action. With hindsight I should perhaps have made the title even less provocative, shortened the Abstract, put the Acknowledgements at the end, rather than on the front page and so on – in each case, “it seemed like a good idea at the time”, and I’m wondering now whether I’ve done something that screams “amateur”. As if that should disqualify me from publishing. But all these things are cosmetic and I was concentrating on the science itself. At the time I didn’t realise there was a significant moderation hurdle on ArXiv – I thought the key was to get an endorser. I was also keen to get my idea “out there”. I figured that if something needed changing they’d just ask me to do it. Anyway, more about my travails in trying to catch the attention of the “physics community” another time.
Here, I want to provide an up to date version of my “Explanation of Apparent Superluminal Velocity in the CERN-OPERA Experiment” (pdf) together with a few words of explanation. This version is significantly different from the one I included in a previous Uncharted Territory post. Postscript: This version has now been submitted to viXra.org, an alternative to arXiv that I came across yesterday.
To recap, my argument is that the neutrinos are travelling at the speed of light, but the speed of light varies slightly depending on direction, because of the motion of the Earth, which is travelling at an estimated 300km/s as measured against the cosmic microwave background (CMB) radiation, which, being the same in all directions, can be taken as providing a stationary reference. It is very difficult to measure the one-way speed of light directly, and the experiments that have been carried out have determined instead c, the “round-trip” light speed. The CERN-OPERA neutrino velocity measurement experiment has unintentionally measured the one-way light speed by comparing neutrino flight times with the expected flight time at c, determined by measuring the distance between CERN and OPERA and successfully transmitting the time at CERN to the OPERA neutrino detector.
The reason for the new version is that I was a bit slow on the uptake. It was only on 27th October, just as I was about to submit to ArXiv, that I belatedly realised that the GPS doesn’t somehow use the “one-way” light speed, but provides a timing based on the “round-trip” light speed. At first I’d thought the whole problem was caused by the procedure for calculating the delay in the optical fibre used to transmit the timing signal over the last 10km into the Gran Sasso mountain to the OPERA neutrino detector. Then, around 25th October, I’d realised that sending a timing signal down an optical fibre runs into the same problems as moving a clock to measure a signal transmission time. On 27th I realised that the GPS also does something similar. A timing signal is sent from the GPS satellite to CERN and to Gran Sasso.
In each case (using the labelling A – CERN, B – Gran Sasso and C – the OPERA neutrino detector) – (i) transmitting a timing signal via GPS from A to B, (ii) moving a clock from B to C to measure a delay in signal transmission, and (iii) transmitting a timing signal via optical fibre cable from B to C – you simply can’t use the time which the transmission has noted it was sent at to measure the speed of light. It’s slippery, but if you really concentrate on the problem, you’ll realise, as did Henri Poincaré, the hero of my previous post, that in each case you have to assume the light speed transmission time.
The best analogy I can come up with is if I received a letter from my nephew – undated, sent 2nd class, with an illegible postmark, as is usual these days – saying I’d forgotten his birthday. I’d be unable to pin down exactly when his birthday was. I’d have to guess how long the letter had taken to reach me.
Similarly, in case (i) in the CERN-OPERA experiment, a “common-view” timing signal is transmitted from a GPS satellite to clocks at CERN (point A) and Gran Sasso (point B), for the express purpose of synchronising those clocks. This signal simply includes the message “the time here is t_SAT”, where t_SAT is whatever the time is on the satellite’s clock. Now, the crucial point (which I only appreciated on 27th October) is that if the message takes n seconds to reach one of the clocks then it’s also n seconds out of date. Unless you know n you can’t tell what the time actually is at the satellite when you receive the message. This is established in the experiment by subtracting the distance from the satellite to the clock and dividing by c, the “round-trip” speed of light, adding the result to the time at which the message was sent. In fact, our whole system of Coordinated Universal Time (UTC) depends on subtracting the assumed transmission time of signals, determined by dividing distance by c, the round-trip light speed. Procedure (i) thus synchronises time between A and B, assuming light travels at c in both directions.
In case (ii) a clock is transported from a master clock at Gran Sasso (point B in the paper, synchronised with CERN by GPS) into the mountain to the OPERA neutrino detector (at point C in the paper) in order to establish the transmission delays in an 8.3km optical fibre (actually this is only the longest fibre in the experiment, but the argument is the same for the others). The transmission time, t_tr in the paper, is the time light would have taken to travel the distance of the cable into the mountain, plus the delays caused by the cable and associated equipment, which I’ve called t_sig. It’s t_sig we want to find out. But again, the signal includes no information about the duration of transmission at light speed. Again, if it takes n seconds to arrive, it’s n seconds out of date when it reaches the clock at the neutrino detector.
Case (ii), though, is slightly different from (i) and highlights the subtleties inherent in the problem. Here, we have a clock at C which we believe shows the time at B, assuming events at B and C are simultaneous. We can therefore establish the delay, t_sig, in transmitting the signal from B to C. But, if events at B and C are not simultaneous, as Einstein suggested, because of the Earth’s motion, then the delay in (or early) arrival of the signal at C compared to light speed transmission from B is exactly matched by the delay in (or early) time at C compared to that at B. Once the clock is moved from B to C it is no longer synchronous with the clock left at B. This is analogous to the Sagnac Effect, whereby clocks have to be adjusted to allow for the rotation of the Earth, and clocks that are moved lose synchronicity. It is in itself an important result, and I intend to devote a post solely to this point (though I don’t always keep my promises). Returning to the CERN-OPERA neutrino velocity measurement experiment, the outcome is that we’re no better off physically moving a clock from B to C than we are transmitting the time from B to C. We always obtain the same delay, t_sig. It might be worth noting that, in the experiment, the same t_sig was obtained by a different procedure (the two-way fibre delay calibration procedure) that doesn’t depend on physically moving clocks.
In case (iii) we do actually transmit a timing signal from B to C along the fibre-optic cables, using the calculation of the delay obtained in procedure (ii). We don’t know how long the timing signal takes, because, again, if it takes n seconds to transmit, it is n seconds out of date, but we assume the delay compared to light speed transmission is t_sig. Thus, the expected time of flight over the entire neutrino flight path has been calibrated based on c, the round-trip light speed – the time of arrival of the timing signal at C against which neutrino flight times are compared is t_A + x_3/c + t_sig (where t_A is the time at A, CERN, when the neutrinos were created and x_3 is, as in the paper, the distance from A to C).
It has to be said that the CERN-OPERA team have done a good job. I’m convinced they are excellent experimenters. The problem is theoretical. It appears they succeeded in establishing a timing signal at point C, the OPERA neutrino detector, that very accurately represented the arrival time of neutrinos emitted from point A, CERN, based on a round-trip light-speed, c, neutrino velocity. The trouble is the neutrinos were only travelling one-way.
The method I’ve adopted in the paper to demonstrate the point is to show that the measurements to determine the expected neutrino flight time at light speed, that is those used in the calculation of the timing signal delay at C relative to A, would all be the same for an observer moving relative to the experiment, but the actual neutrino flight time would depend on the motion of that observer. This is the point of the Lorentz transformations in the paper.
The experimenters have assumed they are stationary relative to the neutrinos, but since the neutrinos arrived earlier than expected, this is clearly not the case. The paper therefore goes on to calculate the velocity of the experimenters relative to the neutrinos, or to be more precise the “frame of reference” of the neutrinos. Because of the details of the experiment this can only be done in the average case. We can only determine one component of our motion relative to the neutrinos, that is that along the Earth’s axis. The rest of our motion varies with the Earth’s rotation and orbit and I assume these motions cancel out to zero.
This is where it gets really interesting. I can’t get the result to tally exactly with the Earth’s motion against the CMB. That leaves me wondering whether the problem is due to experimental error or real – in which case satellites such as WMAP have not measured our motion correctly against the CMB. If the problem is real, and we have measured something other than our motion against the CMB, then things could get very interesting indeed.