Banner
    OPERA Confirms: Neutrinos Travel Faster Than Light!!
    By Tommaso Dorigo | November 17th 2011 07:00 PM | 101 comments | Print | E-mail | Track Comments
    About Tommaso

    I am an experimental particle physicist working with the CMS experiment at CERN. In my spare time I play chess, abuse the piano, and aim my dobson...

    View Tommaso's Profile
    UPDATE: some technical considerations on the measurement are available in a followup post I wrote after attending a seminar on the new result today. In particular, one startling consideration emerges - if the reading of the 20 MHz Opera clock were off by just one tick, the result would be compatible with v=c.

    UPDATE: you can download the new Opera paper at this link. You will need to use the username and password "neuvel".

    The OPERA Collaboration sent to the Cornell Arxiv an updated version of their preprint today, where they summarize the results of their analysis, expanded with additional statistical tests, and including the check performed with 20 additional neutrino interactions they collected in the last few weeks. These few extra timing measurements crucially allow the ruling out of some potential unaccounted sources of systematic uncertainty, notably ones connected to the knowledge of the proton spill time distribution.

    These benefits come from CERN, where proton bunches have been made much shorter: down to three nanosecond long pulses. That means that OPERA can measure the speed of each detected neutrino separately! Of course, with such short pulses, the statistics of protons on target is "only" of 4x10^16, but this is still enough to reach meaningful results from the additional data.
    The figure below show the timing structure of the proton bunches. The black arrow spans 20 nanoseconds to size up the horizontal scale.



    As a sidebar to the improvements yielded by the reduction in the pulse duration, and the related checks that were performed on the source, apparently allowed CERN to spot some less than perfect sinchronisms in the apparatus responsible for the creation of the beam. In particular, the horns that focalize the beam are brought to the right magnetic field by currents that are ramped up before the particles pass through them. It transpired that due to some imperfect arrangement, the currents might still have been ramping up during the passage of the particles, with the result that the focusing of the beam could be less good than predicted.

    This might bring into the Opera speed measurement some systematics due to the fact that the neutrinos produced later -the trailing ones from the spill- woud be better focused (field closer to plateau in the horns). However, this potential issue is made irrelevant by working with very narrow proton pulses (however ramping, the current is practically constant during a very narrow proton spill).

    So what does OPERA find ? Their main result, based on the 15,233 neutrino interactions collected in three years of data taking, is unchanged from the September result. The most interesting part of the new publication is instead that the  find that the 20 new neutrino events (where neutrino speeds are individually measured, as opposed to the combined measurement done with the three-year data published in September) confirm the earlier result: the arrival times appear to occur about 60 nanoseconds before they are expected.

    The figure below, taken from the paper, shows the individual timing measurements of the neutrino interactions from the narrow spills taken between end of October and beginning of November. The red line indicates the average of the 20 measurements.



    It is necessary here to note that since distance from source to detector and time offsets necessary to determine the travel time of neutrinos have not been remeasured, the related systematics (estimated as well as -possibly- underestimated ones) are unchanged. The measurement therefore is only a "partial" confirmation of the earlier result: it is consistent with it, but could be just as wrong as the other.

    Just to make an example, I will reiterate here the doubts I have on one of the time offsets necessary to obtain the timing measurement in Gran Sasso: an 8-km-long light guide brings in a 40,000+-1 ns offset: in order to determine a "delta t" of 60 nanoseconds, a subtraction of that large number has to be made. This offset was measured three years ago, and could have changed if the refraction index had changed even very slightly (e.g. due to aging of the plastic material). This offset was not remeasured in the new analysis, and the possible associated systematic uncertainty remains in my mind an issue.


    One peculiar aspect of the new neutrino interactions is that they highlight the presence of a rather large "jitter" effect: the timing of each neutrino interaction appears to be subject to a "smearing of width of the order of 25 nanoseconds. This could imply that something in the timing measurement is not well under control. Note that such a jitter could not easily be spotted in the global time distribution of the 15,223 neutrinos collected in the last three years; its effect on the global probability density function (which spans 10.5 microseconds) is too small to produce a global shape difference.

    Another note is that OPERA accepted the criticism moved by many on the technique of producing a global probability density function for all proton spills, to compare it then to the observed neutrino timing distribution. They performed a measurement where they analyze the data without adding the functions, and they obtain a consistent result.

    Regardless of all the considerations I made above on systematic uncertainties that might still affect the measurement, I must say I am rather positively impressed by at least a couple of things: first, that Opera has worked like a single man in the attempt of making more solid an already quite scrupulous result; and second, that the OPERA researchers seem now to stand behind the measurement much more united than they were two months ago.

    In other words, those in OPERA who did not sign the first preprint - because they probably did not have enough time to scrutinize all the aspects of the measurement - are now apparently all willing to sign the new one, and probably ready to submit the resulting publication to a scientific magazine. Since I know several members of OPERA and I judge them all serious and scrupulous physicists, their willingness to sign the paper means that we need to take it more seriously than (at least a few among us) have so far.

    UPDATE: I was wrong on that one. It was pointed out to me by a member of Opera that while four of the physicists who had not signed the paper in September now accepted to sign it, four more who had signed the first one now dropped out of the new one! A quite mysterious thing in my opinion, maybe brought by the fact that some (small) mistakes were indeed found in the original analysis (none affecting much the result, however).

    Here is a quote from the conclusions in the Opera article:
    "To exclude possible systematic effects related to the use of the proton waveforms as PDF for the distributions of the neutrino arrival times within the two extractions and to their statistical treatment, a two-week long beam test was recently performed. A dedicated CNGS beam was generated by a purposely setup SPS proton beam. The modified beam consisted of a single extraction including four bunches about 3 ns long (FWHM) separated by 524 ns.

    With an integrated beam intensity of 4×1016 protons on target a total of 20 events were retained, leading to a value of δt measured from the average of the distribution of (62.1 ± 3.7) ns, in agreement with the value of (57.8 ± 7.8) ns obtained with the main analysis. At first order, systematic uncertainties related to the bunched beam operation are equal or smaller than those affecting the result obtained with the nominal CNGS beam."

    And finally:

    "In conclusion, despite the large significance of the measurement reported here and the robustness of the analysis, the potentially great impact of the result motivates the continuation of our studies in order to investigate possible still unknown systematic effects that could explain the observed anomaly. We deliberately do not attempt any theoretical or phenomenological interpretation of the results."

    What will happen next ? Other experiments are already planning equivalent measurements, of course. We owe it to OPERA and its six-sigma result to have opened (or should I say widened, given the previous less exciting measurement by MINOS?) a new avenue of investigations of the physics of neutrinos: a unexpected one, for sure, and probably one which will not bear fruit in the end (but this of course is my personal bias) - but nonetheless some interesting fresh air in neutrino physics!

    UPDATE: Here is the CERN press release:

    Following theOPERA collaboration's presentation at CERN on 23 September, inviting scrutiny of their neutrino time-of-flight measurement from the broader particle physics community, the collaboration has rechecked many aspects of its analysis and taken into account valuable suggestions from a wide range of sources. One key test was to repeat the measurement with very short beam pulses from CERN. This allowed the extraction time of the protons that ultimately lead to the neutrino beam to be measured more precisely.

    The beam sent from CERN consisted of pulses three nanoseconds long separated by up to 524 nanoseconds. Some 20 clean neutrino events were measured at the Gran Sasso Laboratory, and precisely associated with the pulse leaving CERN. This test confirms the accuracy of OPERA's timing measurement, ruling out one potential source of systematic error. The new measurements do not change the initial conclusion. Nevertheless, the observed anomaly in the neutrinos' time of flight from CERN to Gran Sasso still needs further scrutiny and independent measurement before it can be refuted or confirmed.

    On 17 November, the collaboration submitted a paper on this measurement to the peer reviewed journal JHEP. This paper is also available on the ArXiv preprint server


    So, Opera now publishes the results in JHEP. The objection that the result was preliminary is thus off the table....

    Comments

    vongehr
    we need to take it more seriously than (at least a few among us) have so far.
    Finally! Welcome to the dark, sorry, enlightened side. May I humbly point out that the uncertainty is now even smaller (3.7ns) in spite of less events, which once again supports my interpretation of that this smearing is not so much due to statistical error but due to the fact that very large superluminal velocities (V above 10 c) over small initial "jump" distances (x ~ 18m) are involved?
    The 25 ns "jitter" is afflicting the new data but it fits still in with what I said all along. I have this once more clearly explained now here.
    Thor Russell
    Would it make sense now to run the same experiment with much lower energy neutrinos?If your interpretation is correct, then would there be a reasonable chance that the neutrinos would then arrive "on time", not >c. If that did happen, that would appear to rule out many possible systemic errors in the setup also.
    Thor Russell
    vongehr
    I agree that energy dependence is what should be looked at more carefully next; it has been looked at somewhat but not in a way that can settle whether the superluminal velocity V depends on energy or just the threshold to "jumping" [In detail explained with citing OPERA's energy dependence here (end of that article)].

    In short: It is not clear that the energy would change the early arrival time. Why? Well, if the energy just has to get to a certain threshold above which this effect happens, then even the energy's increasing the velocity from 10 c to 10000 c would not change the arrival time. The energy would have to directly change the short initial "jump" distance x in order to show up as energy dependence in the arrival time.

    If x does not depend much on energy [say it is related to the size of a compactified dimension or the distance between two parallel membranes in a stringtheoretical higher dimensional bulk space (see universe on a membrane proposals)], less energy may get you less neutrinos "jumping", but it would not give you energy dependence of the arrival time for those that do "jump".

    Again, all this is the core of the argument of why the low spread in the arrival times is more important than the average of 60ns.
    meopemuk
    Sascha,

    first, I would like to say that I agree with you: The origin of the OPERA anomaly is in the initial instantaneous "jump". The neutrino emerges 18 meters away from the point of the meson decay. However, there is no need to invoke "compactified dimensions" and other exotic ideas in order to explain this unusual behavior. The large distance between the meson decay vertex and the neutrino creation point can be obtained in a simple non-controversial quantum mechanical model http://www.vixra.org/pdf/1110.0052v2.pdf

    Second, if we agree on the "jump" idea, we should be able to predict the outcome of the future (repeated) MINOS experiment. Namely, they will not see any time advance at MINOS, because their setup measures the propagation time between two neutrino detectors, and neutrinos cover that distance with the speed of light.

    Eugene.
    You seem to be confusing the uncertainty in the average arrival time anomaly with the spread in arrival times. Looking at the plot, it's pretty clear that the width of the distribution they see (which ought to be a convolution of the actual distribution in arrival times with the, presumably Gaussian, random error in their arrival time measurements) is something on the order of 20 ns or a little longer.

    vongehr
    No, I do not confuse these. Please see the new post explaining this. The 25 ns are indeed different from the 10 ns.
    And we all know this measurement won't hold up. The hoopla this is generating is disconcerting.

    I cannot understand why are all of you discusing a ghost since the Neutrino doesn't exist because it is a simple mistake by Pauli accepting the SR's equations when there are totaly wrong due that the original equations by Lorentz are also wrong
    Why is Pauli Wrong for Layman
    http://autodynamicslborg.blogspot.com/2011/10/why-is-pauli-wrong-for-lay...
    Calorimetric Experiment (No Neutrino).
    http://autodynamicslborg.blogspot.com/2011/07/calorimeter-experiment-no-...
    Galileo-Newton and Lorentz-Carezani transformation
    http://autodynamicslborg.blogspot.com/2011/06/galileo-newton-and-lorentz...
    Lucy Haye
    wamen. 8ctsirllucyhaye22@gmail.com

    I cannot understand why all of you are discusing a ghost that doesn't exist. Study first, please the following:
    Why is Pauli Wrong for Layman
    http://autodynamicslborg.blogspot.com/2011/10/why-is-pauli-wrong-for-lay...
    Galileo-Newton and Lorentz-Carezani transformation
    http://autodynamicslborg.blogspot.com/2011/06/galileo-newton-and-lorentz...
    Calorimetric Experiment (No Neutrino).
    http://autodynamicslborg.blogspot.com/2011/07/calorimeter-experiment-no-...
    Lucy Haye
    lucyhaye22@gmail.com

    Where did you get the paper? I have been trying to see the v2 but on the arXiv I only get the old 24-long document released in September. The abstract looks different and the comments say that the new file has 32 pages; however, I cannot see the new version.

    If this is true, what are the big implications for physics?

    So just WHERE is this new OPERA paper on the arXiv???

    Bonny Bonobo alias Brat
    So just WHERE is this new OPERA paper on the arXiv???
    http://arxiv.org/abs/1109.4897
    My latest forum article 'Australian Researchers Discover Potential Blue Green Algae Cause & Treatment of Motor Neuron Disease (MND)&(ALS)' Parkinsons's and Alzheimer's can be found at http://www.science20.com/forums/medicine
    CHRIS SCHADT
    Im not a physicist (far from it), but what is the chance that what they have actually measured is the speed of 'spooky action at a distance' and not the speed of neutrinos?
    That is insightful, Christopher. Some people were actually thinking about such things BEFORE the original OPERA result.

    If the faster-than-light number is a systematic error due to a failure to properly account for relativity in a spinning reference frame, then this result is totally expected and still wrong. See Carlo Contadi's letter on the ArXiv: http://xxx.lanl.gov/abs/1109.6160

    Sorry, but Contaldi's assumption is simply wrong: (quote: In an effort to go beyond this accuracy thresh-
    old, the OPERA experiment employed a travelling Time-Transfer Device (TTD) to calibrate the difference in time signals at each receiver. We assume this device to be a transportable atomic clock of sufficient accuracy [15]. The TTD constitutes a classic moving clock synchronisation conundrum in relativity.)

    The TTD is **not** a transportable atomic clock. You can easily chase down the references in the original OPERA paper to understand the TTD.

    matpitka
    Dear Sascha,
     I had in my blog a discussion about the fact that neutrino is produced in a decay of pion meaning that what comes from CERN propagates first as a relativistic pion and then as neutrino. The motivation of the commenter was that the lifetime of pion happens to correspond to length of about 8 meters and he thought that this might not be an accident. Have you considered the possibility of assigning your 18 meter super-fast propagation of neutrino to virtual pion?
    vongehr
    The short initial jump is exactly what makes it possible that the neutrino is actually not to blame at all but instead the pions/kaons while decaying perhaps. I discuss this at the very end of my new post.
    is the "jitter" statistically correlated with deposited energy?

    run a spell checker.

    The Stand-Up Physicist
    The rest mass of the neutrino is not zero, that is why there can be mixing of flavors of neutrinos. I believe the rest mass is all the neutrinos is the most imprecisely known of any subatomic particle (Wolframalfa is reporting 9x10-4 to 3 eV/c2). Let's face it, the rest mass of a particle is one of its most basic properties. The thing I most want to know is how the neutrino gets away with being so darn antisocial, yet still having a rest mass. Guess I will have to wait for an answer.

    Here is a much more statistical question: why with 20 events was the standard deviation lower than with 15,233? The usual pattern is lots of data, smaller deviation.
    I can answer with an example. You get 2 pairs of data, A and B. Here they are:
    A: 3.8,4,4.1 -> Standar deviation= 0.1527525
    B: 1,4,2,6,7,1,4,2,6,7,1,2,4,6,7 -> Standard deviation=2.360387
    The sample B has more measurements but a greater standard deviation (SD). This is because the SD is an estimate of how much the data are distributed around the mean. Consequently, If you get few measurment but close to the mean (that is: more precise data), you will have a small SD. If you have a lot of measurements but with large difference the SD will be greater.

    Please, please, please, change the headline. NOTHING has been confirmed but the lack of a certain systematic error.

    Err. I'm rather miffed, honestly, at this, as well. To my eye, the scientists are saying "it sure LOOKS like FTL neutrinos, so we're going to keep working and make sure we haven't missed any systematic errors." The title of this article, however, seems to assert that they're saying something completely different. In fact, the scientists seem to have explicitly said they weren't going to consider changes to theory yet, whereas the title basically claims that they're doing exactly that. There's a world of difference between "this result isn't due to random error" and "this result proves X, Y, and Z." I find such misrepresentation in a science article pretty frustrating and a bit depressing.

    dorigo
    Hi Exacerangutan,
    Science and journalism are two very different things. This is a science blog and the body of the article talks about science, yet the title is journalistic. Do you find this disturbing ? Too bad. A title a bit over the top helps gain the interest of a wider audience, which is all what science popularization is about.
    Cheers,
    T.
    Tommaso, some people only read headlines. A journalistic approach to presenting scientific findings, even in the title, is a bad idea. I also found this title inappropriate and misleading. Maybe throwing some exclamation points (at LEAST two) on a fact would be a better idea?

    Hfarmer
    Those are people who would not read a more "scientific" title.  
    CERN finds consistent data on faster than light neutrinos.  

    Just would not grab readers and pull them in where they can learn the rest of the science.  If no one who's not already interested reads it, then what's the use?
    Science advances as much by mistakes as by plans.
    So... what you're saying, in essence, is that it's better to completely mislead a large population (but get a lot of attention) than to actually inform a somewhat smaller population (but maintain some scientific and journalistic integrity).

    I'm not sure I can agree with you there, although if that's the standard approach, I can understand why my friend who was studying journalism became soul-crushingly disillusioned with the subject.

    I'm sorry, but if you're going to engage in journalism, and especially to publish something related to science--the expansion and dissemination of human knowledge--I would argue you take on a responsibility to not willfully misrepresent the facts just to create a misguided readership. That isn't science or journalism. That's a frustrated atheist's definition of religion. I don't know how else to explain how disturbing it is--not only that this sort of thing is done, and that there's no sign of embarrassment over it, but that it's vocally defended as appropriate.

    dorigo
    No, I do not feel that title "completely misleads". Opera confirms their original claim with the paper released today. The original claim is that neutrinos are observed to travel faster than light. You might not like the fact, but it remains - Opera's publication is a measurement of the speed of neutrinos, and the results exceed six standard deviation toward the evidence that they travel at v>c. The rest is interpretation.

    Best,
    T.
    Hfarmer
    I will agree with Tomasso that his title is not misleading but for a different reason.  Re running the experiment with the same setup and getting consistent results proves the first result was NOT just a fluke.   Which is the way some treated it.  i.e.  Steven Weinberg 
    The report of this experiment is pretty impressive, but it bothers me that there is plenty of evidence that all sorts of other particles never travel faster than light, while observations of neutrinos are exceptionally difficult.* It is as if someone said that there are fairies in the bottom of their garden, but they can only be seen on dark, foggy nights.


    http://www.scientificamerican.com/article.cfm?id=ftl-neutrinos

    Ouch. 
    Science advances as much by mistakes as by plans.
    Don't get me wrong, the change from "this looks like random error" to "this looks like systematic error" is a big deal. However, the facts remain, Tomasso quoted: "We deliberately do not attempt any theoretical or phenomenological interpretation of the results."

    Contrastingly, Tomasso's headline is "OPERA Confirms: Neutrinos Travel Faster Than Light!!" which is pretty obviously a phenomenological interpretation: it asserts, in short, that special relativity is wrong, which is not actually what OPERA said, unless there was another statement that wasn't quoted here. As far as I can see, all the actual scientists are saying no physical conclusions can really be drawn, yet.

    There's a world of difference between "these are my results" and "this is what I think my results say about reality," and to blur that very, very important distinction is to write misleadingly, whether it's done carelessly or deliberately. The fact that I've had intelligent non-scientist friends convinced that Special Relativity's about to go out the window by headlines like this pretty much leaves no doubt that it misinforms. I don't see how that's in question. Maybe it's just me who considers that a bad thing, though.

    It's already happening on some somewhat scientific oriented forums, and much time is being wasted in clearing the misconceptions headlines like this one create. Asserting that "they wouldn't know better either way" doesn't fly for me either, sorry.

    Yeah, buddy!

    Some people... I mean, "reading the headlines off of arxiv is eye-glaze." It's your story, you tell it; if they don't like it, they can build their own supercollider. ;)

    Oh well, I guess it doesn't matter then if the headline was "The end of the world is coming! ZOMG Ponies!" as long as the article is good. Because everybody knows you can't use the headlines to give you an idea of what the article is about.
    I'm sorry but that doesn't work like that. I'm tired of having to spend time clearing the misconceptions that "mostly harmless" headlines like this one bring about. Yes, I might just let my friends go on and on about how we will be able to travel into the past and kill our respective parents in order to see if a paradox occurs. But there's a difference between doing things like that in jest an in seriousness. Besides, I might even find putting words in my mouth disrespectful. OPERA doesn't assert that neutrinos are FTL. Why should anyone they said so... just because? It brings the reputation of either OPERA or that anyone down. Everyone loses either way.

    dorigo
    Hi Marcos,

    I'm sorry, but "wedo not say that neutrinos travel faster than light" is a childish statement for an Opera scientist to make. Why did they not decide to measure instead something else ? If they publish a paper saying that the arrival time is 60+-10ns before what is expected for a particle traveling at v=c, I do not care if they do not call it "we measure neutrino speeds to exceed the speed of light, and are pretty certain about our result". They did say they are pretty sure about the result, and they did say that neutrinos take less than a photon in vacuum would take to travel the 730km. Too bad if they feel they are misrepresented when one says "they confirm that neutrinos travel faster than light", because that is exactly what their paper says, and arguing against it is childish, period.

    I know that some people are taking this information as a sign that Einstein was wrong, that scientists can't be trusted, etcetera. So what ? There will always be people like that, who use their head to grow hair on top of it. I coulnd't care less if the title of my post makes them happier. I am not misrepresenting what Opera published. Just emphasizing it slightly (like putting exclamation marks in front).

    Cheers,
    T.
    I honestly don't know whether to laugh or cry at the assertion that scientific rigor is childish. On the bright side, though, if this is a typical Science 2.0 writer point of view, I'll know exactly how to respond next time a friend points me at an article here.

    Now, I generally avoid internet arguments, but in the case of someone portraying himself as a science writer and in the same breath scoffing at scientific rigor, it's impossible to keep quiet, so here:

    Even if you aren't a scientist yourself, if you intend to write on the topic with any sort of integrity, it requires a basic level of respect for what you're discussing. The fact that you consider a scientist's skepticism toward his or her own results as childish suggests you have very little respect for how science is actually done. That your initial defense of the headline was on the grounds that you felt it was more likely to garner readers makes you sound like your intent is sensationalism, which is completely the opposite of scientific rigor. This developing pattern is really quite disturbing. I'd say I hope it's a fluke, but that your response is 100% defensive with no hint that you see anything you might change in the future makes it seem unlikely.

    By the reasoning you've presented, you would have been proud to write "RUTHERFORD PROVES: ALL MATTER UNSTABLE!!!" following the publication of the Rutherford model of the atom. Science is at least 50% about being wrong, and very few measurements are direct or simple. What's childish is flying off the handle and making grandiose, sensational claims.

    Tommaso is a working scientist, and I'm a naive philosopher; neither of us are saying anything of the sort. Are you trying to say, if you were Rutherford, you wouldn't be excited? You wouldn't write in your journal, "All matter unstable!" because you were thrilled to be the one to whom the universe whispered this secret?

    Then you'd be Paul Dirac, who didn't believe what the equations were telling him. He thought he made a mistake, somehow stumbled upon the proton - and he felt kind of silly about it - until the positron was validated in a cloud chamber.

    Both are the ways of science. If one is a cold logician, one might write in a dry, technical style - that attracts only the dry, technical types - or one might be passionate, like Einstein, who spent the second half of his life applying skepticism to quantum mechanics with verve and aplomb. Was Einstein a scientist?

    You are correct in your skepticism; you are incorrect to consider science to be able to function without the human element - the scientist.

    Hi Tommaso,

    I think it`s a funny title, but You certainly know what Your blogging friend would say about it ... ;-)

    THIS WEAK HYPE

    ;-P

    dorigo
    Don't share the view, it's not hype. It's a honest measurement which has undergone more scrutiny than 99% of other results that get routinely published. That does not mean it's correct, but it is certainly worth our attention. And it does imply that neutrinos are superluminal, like it or not.

    Cheers,
    T.
    has the effect (geological) of continental drift been accounted for? if the italian peninsula is moving slowly toward the alps, then of course the neutrinos would appear to travel faster than light, because the distance is changing (getting smaller)

    Hfarmer
    As others have said, and you yourself pointed out, the latest numbers from CERN are still subject to the same systematic errors which have been pointed out already.  Given those errors it is not surprising that they got a consistent result.    
    What has been confirmed is that the scientist at CERN are at least competent experimentalist who are not making utterly random errors.  They are self consistent.  No one ever said otherwise (competent and even "great" scientist can be and are wrong every day.) 

    I will be convinced when Fermilab, which is independent, and using a somewhat different setup confirms this.  Until then I stand by what I had to say about this months ago.  
    Science advances as much by mistakes as by plans.
    The original OPERA pre-print had 16,111 neutrino events, version 2 uses 15223 events. Any idea why they deselected 888 events?

    dorigo
    Yes. They discarded 5% of the statistics which they found out had flaws (was in part compatible with electronic noise etcetera).
    This, and other unpleasant details, have emerged during the more careful reanalysis. But the main result stays...

    Cheers,
    T.
    "Science and journalism are two very different things. This is a science blog and the body of the article talks about science, yet the title is journalistic."
    Yes its true...

    The headline (and some similarly aligned statements in your text) are misleading. OPERA cannot confirm its own measurement; you point this out, but then qualify it by labeling it a "partial confirmation." Its not. It's a refinement through the adding of more (possibly better-collected) data. ("Data," not "statistics": statistics are abstractions from data. The measuring of 20 more neutrino events is the collection of more _data_. A new calculation of the time offset from this new data is a new _statistic_. Physicists need to start using these words correctly.)

    That aside: Has OPERA released a plot of the time offset vs energy for the new events? Is there anything to see in that or is the energy spread too small?

    dorigo
    Sorry, I think you did not pay too much attention to the new analysis. The reanalysis which uses separate templates in the likelihood (so not using more data, but using a different test statistics), and the use of narrow-pulsed protons (new data), together allow to rule out some possibly unaccounted sources of systematic uncertainty -mostly ones connected to the source and to the statistical analysis. In this respect the new result is a confirmation: it not only is more precise; it also is much less criticizable.
    I do view the narrow-pulsed proton result as a different experiment, at least in part.

    About the wording, I concur. But you need to realize that here I am not talking to statisticians or physicists, but I am trying to do science outreach. I am much more careful with words when I serve as a member of the CMS statistics committee, for instance... ;-)

    Best,
    T.
    Hank
    I don't think the wording is out of place.  I am harder on media exaggeration than anyone here - my ridicule of New Scientist for going out of their way to craft controversy in their titles could take up a whole book - but it was expected they would revisit this and they did.

    This is an interesting result, and an exciting one, partially because it reminds the public there is more to physics than the Higgs.
    Confirmation doesn't come through analysis techniques, it comes from new independent data sets---separate experiments. I agree that the newly taken data _bolsters_ the original claim, but it suffers from the identical problems as the original data set: measurement of the distance and signal delay.

    The reanalysis of the original data set is certainly not a confirmation---it's just an improvement of the original result. Why they used the averaged-beam-pulse analysis in the first place, rather than an event-by-event analysis---I don't understand that.

    I'm no scientist but here what confuses me, and I'm sure the explanation is simple. If moving faster than the speed of light is not possible, how does the universe expand faster than the speed of light? Is it just a combination of out speed combined with another galaxy moving away from us?

    The difference is that nothing with mass can move faster than the speed of light. When you talk about the universe expanding faster than the speed of light, it is space itself that is expanding, not a particle with mass.

    Hank
    Indeed, so if someone says 'nothing can move faster than light', it means quite literally nothing - space.  This is why the universe can be double digits in billions of years old and triple digit in billions of light years in size.
    Hmm, ok. I'm obviously not a physicist, but I thought light doesn't have a mass either and I assume light can't travel faster than light? Why can space than travel faster?

    And to use the opportunity, I was always wondering that if the space expands (the universe gains m³) and vacuum has an energy (if I remember correctly) does this mean that the vacuum energy per m³ decreases over time so that the overall energy in the universe remains the same?

    Thanks for an interesting article, discussion and making an effort in explaining these complicated things to interested lay people.

    Quick and most certainly incomplete answer to the light question: E=mc^2. So yes, light has mass... in a certain way, since it has (is?) energy :) I think you would need to be able to stop light to measure it mass though. I have no news of anybody having done it yet.

    Halliday
    Anonymous:

    In the case of "the universe expanding faster than the speed of light", you correctly state that "it is space itself that is expanding".  Unfortunately, you then add "not a particle with mass", as if you were talking about something moving (as in "move faster than the speed of light").

    The point is that the amount of space between objects (galaxies, etc.) is expanding, increasing, "stretching", etc.  Space itself is not "moving" anywhere, especially not "through space".

    David
    I see, nice. :D

    A simple explanation of OPERA results without strange physics
    http://arxiv.org/abs/1110.0239

    Mr Henri, I find your explanation very interesting and moreover much more simpler and convincing than anything I have read about it. Indeed I can't tell if your idea is relevant to this faster than light case: It needs experts of the field to discuss it but to a mere layman like me it's seems definetly a smart idea. Jean-Pierre

    I don't know about the content of this report. I thought it was cool so I clicked the link. Then while waiting for it to load in another tab, I got blasted with a loud commercial. Then when I tried to click the obvious sound button, it didn't do anything. Then i noticed another smaller mute button in the opposite corner of the ad. So I clicked it.

    THEN, I scrolled down here to comment on the inappropriateness of having an ad whose audio is opt-out, rather than opt-in, and while I was typing, I got blasted with ANOTHER commercial from my speakers from the same ad that I'd just muted.

    I don't care what your wonderful new scientific discovery or confirmation is, anymore. F*** this website, and any others whose advertisers don't know how to respect their readers.

    Antonio Erediato just stated, via the New York Times, all the researchers have signed on to the latest paper. No dissenters. The whole news about people dropping out after such a remarkable "repeat" of the result seemed odd.
    http://www.nytimes.com/2011/11/19/science/space/neutrino-finding-is-conf...

    Tommasso is there any discussion on reducing the energy of the nuetrino production with the same test. If the problem is systemic or Sascha pions the number wont change. If it's in the actual nuetrino travel section one would well expect a change.

    Doesn't the smearing of arrival times depend on the size of the detector? I think it was quite a big thing that detector. Some neutrinos go deeper in it before interacting.

    dorigo
    Hi Eirik,
    no. The reason is that the time which is taken is the one of the muons crossing the same layer of detector (orthogonal to the arrival direction). Since these muons can be produced up to 300 meters upstream (when neutrinos hit the rock before the detector, rather than the lead of the target of Opera), what is measured by Opera is actually a combined travel time of neutrinos AND resulting muons. The latter are relativistic and travel at the speed of light.

    In any case, Opera also operates a correction due to the angle which the muons make with the nominal direction between CNGS and the detector. In other words, if a neutrino-rock interaction 200 meters upstream occurred 200 meters off the nominal trajectory (neutrinos fan out as they leave CERN, and have a spread of about 1.5 km transversally as they reach Gran Sasso), the muon which is later detected in Opera would have traveled not 200 m, but 282. This is taken into account by the analysis.

    Cheers,
    T.
    I'm by no means a scientist but has general relativity been taken into account? CERN and Gran Sasso have an altitude difference of around 2500m! This means that, according to Einsteins theory of general relatvity (which if memory serves does NOT rely on a maximum speed) time will be slightly dilated at CERN. A better experiment would involve a new sample of neutrinos being released from Gran Sasso when the first ones were detected. This would remove the potential error of having two clocks, although would not prevent this time dilation effect from being observed.

    I don't know the magnitude of this affect though, and it could have been taken into consideration.

    dorigo
    Hi Anon,

    yes, the effect you mention has been taken into account (it is rather minuscule anyway).
    Cheers,
    T.
    The only interpretation for the OPERA experiment is that
    http://vixra.org/pdf/1111.0001v1.pdf

    I thought they were disputing this over a satellite glitch. I still believe the problem with making the speed of light a speed limit is that it is based on an axiom that I believe is wrong. The Lorentz Transformation he used has an observer frame that is fixed. Even as an event occurs in another frame the observer frame’s time doesn’t move forward. This seems counter intuitive and against any idea of relativity. I also think he may have muffed it on gravitational lensing. http://jetsrock.wordpress.com/

    Does this make them faster than the air speed velocity of an unladen African Swallow?

    I don't know-AAAAAAAAaaaaaaaaahhh!

    As a theoretical physicist, I prefer to wait MINOS confirmation 2012, before any interpretation. I don't know, why media spread OPERA data so quickly?

    As I wrote in another post, Z=117 element discovery was waiting about a year, before any news outside scientific community. Why this way is not applied in the OPERA case?

    This is not conservatism. This is how science should work. Our job is not like a Hollywood entertainment.

    Cheers,

    Nick

    dorigo
    Hi Nick,
    the problem is that without the money that comes from the average Joe, our science would be impossible. We are paid to produce results and have the moral obligation of getting them out as soon as possible. As a taxpayer I am pissed off when an experiment I have no access off "holds on their data" for long.
    Cheers,
    T.
    In 2008 there was a sensational article in Nature about super-allowed proton decays from high-spin states in few nuclei. This was an experiment at GSI in Germany. As you can figure out, this result contradicts with the relation between angular momentum and the centrifugal barrier. The higher the angular momentum, the higher the centrifugal barrier, which means proton decay can't be super-allowed at high spins, if there is nothing else, except protons and gamma rays. The GSI result was so shocking that some theorists started speculations for new exotic particles from nuclei. However, an year latter MSU tried to repead the GSI experiment and MSU didn't get the same result. Moreover, MSU found an error into the GSI experiment related with the coincidence between protons and gamma decays. This is an nice example that sensational result can be a risky stuff. Of course, there are many other examples like this.
    The contradiction between OPERA and ICARUS reminds me the GSI-MSU saga.

    dorigo
    Indeed, the likelihood of a mistake is proportional to the number of standard deviations of a result from the expectation...

    Cheers,
    T.
    dorigo
    Or more precisely (since this thread is on a statistics article), the probability that a revolutionary result is mistaken is roughly equal to (1-p), where p is the estimated p-value of the data given the current theory.
    Cheers,
    T.
    A question I have is how the established velocity of light in vacuum has been measured. It seems that the definitive measurement is with lasers by Evanson et al ( http://rationalwiki.org/wiki/Speed_of_light ) but I have not found a reference or link. I would be interested to see if the experiment was done in vacuum. From a description I found, it does not seem that it was.

    They give a value with an error of 0.001% in km/sec. Now the atmosphere is full of electromagnetic waves, from radio to light to infrared so it is not exactly a vacuum. When one reaches such accuracies maybe even the magnetic field of the earth, let alone any static electric fields or magnetic materials in a lab, might make a difference. I was wondering whether a beam line at CERN could be used with its vacuum and shielding for beam stability to redo the experiment.

    In a sense it might be that the neutrino velocity they are measuring in OPERA is a lower limit of the true velocity of light in vacuum, as neutrinos are so weakly interacting. The jitter also could be due to two body elastic scatters every odd neutrino or so, over the large distance traveled.

    dorigo
    Hi Anna,

    forget the jitter, it is due to the clock in Opera working at 20 MHz -which means that their time measurements have an accuracy of +-25ns, with a flat distribution.

    As for the rest, I do not know about whether beam lines at CERN could be of help. But the speed of light in vacuum is a constant of nature and the metre is defined from its definition. But before this, c was known to a few parts per billion, so I do not think we need to worry about this.

    Cheers,
    T.
    Icarus's team is only confirming the scientific truth with the effect Cherenkov, which plans effectively; that a particle going faster than the light to a defined middle (or the rate of the light maybe lower in 300.000Km/), a part of its energy loses by emitting pairs of electrons-positrons etc.....
    But the objective which should have this team would be rather the research for why of these 60ns, the team of Opera of which measured well and truly and which is an effective and real measure.
    No current theory plans a new phenomenon, which could explain it. Thus only a new approach of the physics can take out us of this uncertainty, without putting in defect the relativity of Einstein.
    Obviously the speed of light in the space is, and will be for a long time a borderline speed, moreover the fundamental elementary energy of particles requires it.
    Conclusion: Icarus's team is irrelevant.....
    A C E

    OPERATIC CONCERN
    -- James Ph. Kotsybar

    Oh, little neutral one of tiny mass,
    who flies anomolously from the sun,
    you zip through matter photons cannot pass:
    Could this explain the races you have won?

    From Einstein, few believe that it could be
    that any mass can go as fast as light --
    it’s deemed complete impossibility,
    assuming Relativity is right.

    If proved, the implications terrible,
    will give complacent physicists a scare.
    In terms that twist the ancient parable
    it’s you that’s tortoise; the photon’s the hare.

    It seems, though steady, light can’t keep up pace.
    You oscillate, and yet you win the race.

    I'm a laymen. In the excitement I downloaded the 09/22/11 OPERA paper (the first experiment) when I found the link to it on a blog. I did bust my hump trying to interpret it. I won't make that mistake again only to wind up posting embarrassing layman questions on a science blog.

    But anyway, here goes -- this section (among others) has me stumped:

    "The time of flight of CNGS neutrinos (TOF ν ) cannot be precisely measured at the single interaction level since any proton in the 10.5 µs extraction time may produce the neutrino detected by OPERA."

    Does this mean that a neutrino that arrived at Italy from Switzerland may have been produced anywhere within the CNGS apparatus or between those locations? Is it important for calculating the neutrino's speed exactly where the neutrino was produced?

    Does this mean they must do some math to correct for it?:

    "However, by measuring the time distributions of protons for each extraction for which neutrino interactions are observed in the detector, and summing them together, after proper normalisation one obtains the probability density function (PDF) of the time of emission of the neutrinos within the duration of extraction."

    Does it matter? Thanks.

    I am a laymen and not a scientist. This is from the 09/22/11 paper (first OPERA experiment):

    "The time of flight of CNGS neutrinos (TOF ν ) cannot be precisely measured at the single interaction level since any proton in the 10.5 µs extraction time may produce the neutrino detected by OPERA."

    I'm looking at Fig. 1. Is it correct to interpret the statement above to mean that it is not clear exactly which specific neutrinos produced between TT41 and the Hadron stop over some period of time can be matched with the neutrinos detected at LNGS? Is this why the duration of the burst was reduced in the second experiment?

    Is it correct to interpret the following to mean that some math was done to correct for the conditions described?

    "However, by measuring the time distributions of protons for each extraction for which neutrino interactions are observed in the detector, and summing them together, after proper normalisation one obtains the probability density function (PDF) of the time of emission of the neutrinos within the duration of extraction.

    Thank you.

    dorigo
    Yes, a proton spill in the original experiment lasted 10 microseconds. This produced many neutrinos along the path, and each of these could be produced at any time during those 10 microseconds. However their probability density function over the 10 microseconds is expected to match the proton spill structure. This is what their stat analysis assumed in the first measurement.
    Cheers,
    T.
    Could the neutrinos be influenced by dark flow/dark energy (we Know black holes can bend light but how far does there influence extend) i.e. to all matter in the universe ?

    dorigo
    Hi Colin,

    how would I know ? This looks like science fiction to me...

    Cheers,
    T.
    Dear Mr. Dorigo,
    I would like to ask a clarification

    Opera-CERN experiment seems to have many inherent problems, namely:

    -neutrino 's speed measurement has-been made ​​one-way,
    therefore leading to timing sinchronisation problems

    - FTL-hypotesys contradicts SN1987 measurement (although different neutrino's energy)

    -FTL-hypotesys is not accordlying with energy measurement by ICARUS experiment

    -Experiment has been made by measurements throug opaque rocks,
    hence excluding the possibility of direct light-neutrino speed comparaison

    - neutrino's speed is measured to exceed "c" value of only 0,001%
    and it appears quite "suspect".
    In fact, If neutrinos were really FTL, we could expect they were able to CLEARLY exceed "c"

    Resuming, do you think It would be possible to make experiment in free air, directly comparing neutrino and light speed?
    For example, it seems there is a neutrino detector located at Mont Blanc (already used in SN1987 detection)
    It would be possible to fire both light and neutrino beam throug free atmosphere from Geneve to Mont Blanc?

    dorigo
    Hi Roberto,

    I don't think so. Photons in air do not travel at the nominal light speed either; furthermore, there is no way one can produce a beam of neutrinos and light in synchronism with each other.

    There are better ways to test the result, and they are being tried.

    Cheers,
    T.
    Three authors who signed the first paper, are missing in the final paper
    A. Schembri,
    P. Strolin
    D. Naumov,

    http://arxiv.org/abs/1109.4897v1
    http://arxiv.org/abs/1109.4897v2

    Can someone explain this?

    Regards,

    dorigo
    Hi Titus,

    quite simply, they did not sign. Any author has such a choice. One might suspect that they were in disagreement with some aspect of the data analysis, but it remains a speculation.

    Cheers,
    T.
    Dear Mr Dorigo I thank you very much for your answer

    Final consideration: if CERN results were correct, repeating the experiment across a longer distance it should give a proportionally bigger Dtime
    ("Dtime" is the presumed neutrinos anomaly early time arrival)

    For example, being the actual emitter-to-detector (CERN-to-GranSasso) distance about 730 km ad Dtime= 60ns,
    using a new beam path 7300km long they should get Dtime=600ns

    So, to validate results, CERN could fire the neutrino beam toward Japan or USA neutrinos detectors (very far from Geneva)
    and Dtime should change accordlying to the distance increase

    Furthermore, using long distances would increase a lot Dtime value,
    overwriting any expermental error issue

    What do you think about ?

    Thanks in advance

    Roberto

    dorigo
    Dear anon,

    your assumption rests on a model which Sascha (here at alpha meme) would object to - read his postings on the matter. I do not support his suggestion that neutrinos first travel at multiples of c for a very short time and then slow down or whatever, but this is just to tell you that whatever you do only tests something according to some hypothesis.
    In any case a beam of neutrinos is not something you can move around as you please. There's a complex infrastructure that took years to build, which produces neutrinos pointing to LNGS. And if they could move it somewhere else (they can't without investing O(50 M$) in it or more) the neutrinos by traveling for 10x the distance would defocus by x10, producing a beam 100x less intense in Japan. That means that a Opera-like experiment there would detect only a few neutrinos per year. Not a good idea.

    Cheers,
    T.

    The particles of OPERA experiment (faster than light) may be the non-conventional particle(NC Particle) defined by "Double Relativity Effect". But NC particle is explained by Theory of relativity.The particle will be similar to photon in all respects. only quantity will be deffer.It is not a particle of this space time.

    I would like to point out that besides the relativistic corrections considered by van Elburg on the measurement of the distance from source to detector, and by Costaldi on the influence of gravitation on the synchronization of the clocks, there is also another relativistic effect which does not seem to have been taken into account. It stems from the impossibility of permanently synchronizing clocks co-moving with a rotating body (the Earth in this case).
    In the framework of Special Relativity (i.e. neglecting gravitational effects) let us consider the inertial reference frame with respect to which the centre of the Earth is at rest (exactly initially, and with good approximation for a period of, say, 24 hours). In comparison with the standard time T shown by a system of mutually synchronized standard clocks at rest with respect to this non-rotating frame, any standard clock co-moving with the surface of the Earth at a given latitude is slower: if initially it had been instantaneously synchronized to the clocks of the inertial frame,, at standard time T it shows the time γT, where the contraction factor γ decreases with the latitude in the northern hemisphere. Since the latitude of the OPERA laboratory where the neutrinos of the experiment are detected is lower than the latitude of the CERN laboratory near Geneva where the particles are emitted, a clock at the detection site initially instantaneously synchronized to an identical clock at the emission site gradually builds up a delay with respect to the latter, at a rate which turns out to be of the order of 1.5 nanoseconds per hour of standard time. Therefore the time of flight of the particles measured by the laboratory clocks appears smaller than the time of flight seen from the inertial frame, by an amount proportional to the time interval separating the actual observation of the flight from the last synchronization of the clocks preceding the observation. The average value of such time intervals over the observations performed in the experiment is an indispensable information (which does not seem immediately extractible from the data communicated) in order to estimate whether the correction associated with this relativistic effect is negligible or not.
    vc

    Hi, Mr Tommaso Dorigo,

    Many people think that the synchronization of clocks is the problem of measuring the speed of neutrinos in the experiment Opera.
    A paper is currently submitted for publication shows that this time of 60ns is consistent with the effect Shapiro.
    I ask Mr Tommaso Dorigo, a few comments on this article, and perhaps pass it on to Mr Autiero, head of the Opera experience, in fact he does not answer my emails why?
    Thank you in advance for your answer. The title of this article is:
    Additional delay in the Common View GPS Time Transfert, and the
    consequence for the Measurement of the neutrino velocity with the OPERA detector in the CNGS beam [
    you can read in :
    :www.new-atomic-physics.com
    A C E
    nap01

    As a non-scientist, something about this proposition perplexes me, but I'm not sure exactly what it is. Are clocks synchronized differently when measuring the speed of photons? In other words, if using the same timing devices used in the OPERA experiments, would photons be found to be traveling at anything but their expected speed?

    As a non-scientist, something about this proposition perplexes me, but I'm not sure exactly what it is. Are clocks synchronized differently when measuring the speed of photons? In other words, if using the same timing devices used in the OPERA experiments, would photons be found to be traveling at anything but their expected speed?

    If we replace neutrinos by photons, propagating in a tube empty of air in the experiment Opera, we would have the same result . We would find that these photons go faster than the light, it would be a nonsense.
    In this article nap01 show that it is necessary to take into account a delay Shapiro which is not the same for the stations of the CERN and the Gran Sasso. This delay is not at present recorded in the experiment Opera.
    ACE
    nap01

    Thank you very much for your reply. Well, it is your argument to make and someone with more knowledge than I on these matters to affirm or reject.

    In a non-inertial reference frame which does not admit a permanent synchronization of co-moving clocks, such as the Earth crust, the very definition of "velocity of light (or photons)" between distant events poses problems and crucially depends on the (non-unique) convention adopted to set the clocks. The invariance of the velocity of light only makes sense and applies to inertial (or locally inertial) frames. So, just like for the neutrinos in the OPERA experiment, the "velocity" of the photons as conventionally evaluated in your thought experiment would indeed be greater than the velocity of light with respect to inertial or locally inertial frames, but there is no nonsense in this.
    The Shapiro delay does have an effect , but a numerical estimate shows that this effect ifalls by many orders of magnitude below observability in the OPERA experiment.

    The claim is challenged. So, are you saying that we don't have the same conditions posed here - with the rotating earth and moving satelite situation, as the special conditions described under which we would find photons traveling at expected speed. We either need the entire testing apparatus to be in the same reference frame, use some other method of time-keeping, or expect the irregular results cited here. The effect of the phenomenon on OPERA's results is negligable irregardless. Thanks in advance so that I don't have to waste space just to make a "thanks" post.

    dorigo
    Well, I do not know what to answer, so I'll just cash the thanks...
    Cheers,
    T.
    BDOA
    Azzum doesn't have the only explanation (and i think that one is poor, invoking quantum tunnelling where the energy of the neutrino is already positive and large). I already have another explaination of the OPERA measurement, which is standard QFT plus the Scharnhurst effect and neutrinos having guage field to interact with. Modifying special relativity for one result, is a very big step to make. Another explanation already in the literature, is Double special relavity, usually leading to a speed up linear or v^2 in energy.

    If OPERA is now confirmed, (and I can't see textbooks changing relavity without a several other experiments.)

     to resolve what explaination to use for it, we need measurement of the change in velocity versus particle energy, versus neutrino type, and versus the material between CERN and OPERA.

    Scharnhurst+neutrinos interacting with a field, the velocity varies with the density of the material only.

    In DSR, the velocity of neutrinos varies will neutrino energy only.

    In Theories based on neutrino mixing the speed with primary vary with the type of neutrino.
    BDOA Adams, Axitronics
    Thing about systematic errors is that they hit you from behind. One possibility that comes so far from behind you might think its too boring to even look at, is the contribution to baseline length by the survey along the Gran Sasso tunnel. The paper dealing with that is Opera Public Note 132. There are three really surprising features of that report, that seem to invite errors that could bend the tunnel survey and so change the baseline length from CERN- that baseline being nearly at right angles to the general tunnel direction.

    (a) The shortness of the separation of the GPS control points used to bring orientation into the tunnel. (Table 1) At the one end GPS1 and GPS2 are only 50m apart and at the other end GPS3 and GPS4 are only 80m. The relative position of close GPS points can easily be in error by a couple of millimetres, so these paired orientation points should each have been a few kilometers apart.

    (b) The asymmetry of the survey method (Figure 2) bearing in mind the danger of horizontal refraction bending the traverse rays. That systematic effect threatens when a traverse line is near one wall of a tunnel and there is a temperature gradient away from that wall. The main traverse was close to one wall, supported by one-directional sightings onto reflectors on the other wall.

    (c) The lack of supporting gyro observations. A gyro theodolite can control absolute directions underground, thanks to Earth rotation. Although their set up allowed for future gyro measurements the surveyors reported simply that they didn't use them.

    The most well known case where gyro measurements prevented a snafu was in the survey of the Eurotunnel from the English side. This was reported by Korittke,N (nd) "Influence of horizontal refraction on the traverse measurements in tunnels with small diameters ".

    www.slac.stanford.edu/econf/C9009106/papers/023.PDF

    After only 3.6km of traverse the gyro picked up an accumulated error of 1m. That was in spite of the traverse survey being a symmetrical zig-zag down the tunnel, specifically aimed at cancelling horizontal refraction effects. A main contributor was thought to be a slight curve in the tunnel introducing an asymmetry. The Gran Sasso tunnel has a major and long curve in it.

    I can't myself see horizontal refraction being able to account for an 18m error in the baseline length due to bending in the direction of a 10km tunnel. Perhaps a couple of metres, if the error were really scandalous. But my point is this. Suppose all the authors were scurrying around to find possible systematic errors that could change their result. Then what the surveyors should have been scurrying to find was evidence of horizontal refraction. The first check, as elegant and economic as Eratosthenes measurement of Earth radius, would be to treat their survey as an as-built measurement and compare it with the original design geometry or the tunnel, done by surveyors whose jobs had been truly on the line and who had proved their correctness by accurately holing through. If that revealed no problem the next step would be to hire a gyro theodolite and take some measurements at the portals.