The New OPERA faster-than-light Neutrino Test: Results!

Q: “Why don’t physicists shield themselves from neutrinos?”

A: “Because they never see them coming.” #neutrinojokes

Over the past two months, we’ve talked more about neutrinos than ever before thanks to an extraordinary claim that neutrinos have been observed to move faster-than-light!

And as you well know, no particle is allowed to travel through spacetime faster than the speed of light in vacuum, no matter how much energy you put into it!

(Unless, that is, you count what happens in video games.)

Here’s a brief refresher for you as to what’s been going on. CERN, the home of the world’s most powerful particle accelerator, has (as sort of a side-project) been producing a beamline of high-energy protons, and firing them in a very specific direction.

Image credit: The OPERA collaboration's recent paper.

What happens to these protons? They smash them into a “fixed target,” and the protons are moving at such high energies that they can produce all sorts of extremely high energy, unstable particles!

More specifically, the vast majority of these particles wind up moving at somewhere around 99.999% the speed of light, with a good number of them decaying into muons and — to be specific — muon neutrinos.

These neutrinos then travel through the Earth for about 732 km, before winding up at the OPERA detector beneath the mountain of Gran Sasso in Italy. The very interesting thing, of course, is that the OPERA team found that the neutrinos arrived about 60 nanoseconds too early!

What basically happened is they created a series of large pulses of muons and neutrinos over a time period of a couple of years. They calculated, based on their measurement of the distance between the source and the detector and their understanding of their electronics, just how long it should have taken the neutrinos to arrive at the detector.

What they determine is that, if you take the light-travel-time between the source and the detector, then factor in a 986 nanosecond delay for all of the known factors that can affect the arrival time (e.g., the electronics, GPS timing, effects of general relativity), you should get a perfect match of the detected neutrino signal with the source signal.

But they didn’t! They found that you need a 1043 nanosecond delay, for a surprising result that the neutrinos arrive around 60 nanoseconds early. If this is true, it means that the neutrinos traveled just a tiny bit faster than what we think of as the speed limit of the Universe: the speed-of-light in a vacuum. How much faster? Just 0.002% faster than light.

Like many others, I immediately speculated that this team was fooling themselves with some sort of error, spoke publicly about it, and kept reporting on the latest tests and developments. (Even my car got in on the action.)

After all, their delay was based on the data shown in the above graphs. Is it really compelling that you couldn’t shift that graph by 50 or 60 nanoseconds — just one of those bins — and still get a very good fit? Hardly.

So, they did a very clever re-test, which would basically test whether my ideas numbered 2 or 3 were the culprits. The results are in a new release of their paper (pdf), which is hard to find (the arxiv show only the older version), so the link to download it is a copy that I directly uploaded to this site. Other very interesting takes and writeups are available from Chad Orzel, Matt Strassler, Tommaso Dorigo (who has some objections, all of which are very weak) and Sascha Vongehr, as well as even making mainstream media news! Here’s what they did.

(All subsequent images taken from this new version.)

Instead of creating one long pulse, and then reconstructing the shape of that pulse at the end, they instead sent a series of very short, three nanosecond pulses of high-energy protons instead. Moreover, they spaced these pulses out by some large amount of time — over 500 nanoseconds — that’s much longer than the uncertainty in the arrival time. This way, when you observe a neutrino in the detector, you can know definitively which pulse of protons it came from!

Again, these neutrinos were expected to arrive in the detector at the light travel time, offset by 986 nanoseconds. That corresponds to the “zero” mark on the graph below. Where did they show up?


Even with this new procedure, with these extremely short pulses, the neutrinos still all come in earlier than expected. So there’s no detection bias, which means that their earlier results are positively confirmed.

We can also take a look at their statistical error, which is a bit unusual.

Normally, you’d see a bell-shaped curve in the distribution of the arriving neutrinos. What we instead see is flat. OPERA, however, explains that there’s an observable jitter of ±25 nanoseconds due to “the tagging of the external GPS signal by the OPERA master clock.”

This was glossed over in the original version of their paper, but knowing this now, it totally explains why they put their data into 50 nanosecond bins! Now, it’s worth pointing out that a 50 nanosecond shift in the right direction would all but nullify this result.

And that’s the last major skeptical argument for what’s going on here. It is conceivable that they’ve got a systematic error in their expected delay calculations, which may be due to something like the atomic clocks, the measurement of the baseline distance, an electronics triggering mishap, or some other mundane reason akin to these.

But if there isn’t anything like that, then neutrinos really are traveling faster than we expect them to!

Now, let me ask you this: if you’re a good scientist, what’s the next thing you’d demand?

Image credit: The MINOS/NuMI collaboration with Fermilab.

You’d independently check the results by performing an even more precise test of this exact effect, and try to reproduce it! That’s exactly what MINOS is going to do over the next two years, and that’s going to be a very interesting result!

Because if they disagree, we’ll be able to say that one team likely made a mistake. But it’s highly unlikely that both teams will make exactly the same unexpected, unaccounted-for mistake, and so MINOS may very well confirm that these neutrinos are, in fact, moving faster than they ought to!

Image credit: NASA, ESA, K. France, P. Challis and R. Kirshner.

In which case, we’re going to have a brilliant, legitimate theoretical conundrum on our hands, trying to reconcile why neutrinos from different sources (like supernovae, particle accelerators, nuclear reactors, etc.) appear to travel at significantly different speeds from one another! Some people are speculating already; I’m going to exercise a little patience for now.

But this is a very interesting time to follow what’s going on in the world of neutrinos, and the next round of experiments will either confirm OPERA’s bold findings, or they will wind up with a fair amount of egg on their faces…

47 thoughts on “The New OPERA faster-than-light Neutrino Test: Results!

  1. If what they’re doing is timing the arrival of the neutrinos, shouldn’t they have used a different clock? It looks like they used the exact same method as before.

  2. They are using the same method as before; the only change was in the timing of the proton pulses generating the neutrino beams. Personally, I think it’s the right call — it minimizes the amount of re-analysis that has to be done and is more directly comparable to their original results than if they had made a lot of changes.

    The real test will be to see what MINOS and T2K have to say.

    Dr. Dorigo notes that a systemic one-tick error in the 20MHz timing clock would just about completely eliminate the measured offset. He also noted surprise about finding out about the 20MHz clock now. Has the analysis done by OPERA ruled out this potential error?

  3. I’m still a little bit put off by the spread, and total counts/bin (as high as 2!), but… their method was a wonderfully simple way of removing a difficult to analyze source of error. What would be really interesting is if that bare hint of a double peak structure ends up being real. I still think things will come down to some lag time that wasn’t accounted for, and my money is on an IC or two having a respin without notification. It sounds like their clock system sucks if it has 25ns jitter,

  4. Very nice follow up work by OPERA folks and now we just have to wait. Speculation doesn’t have to wait; but on this one I’ll keep my personal speculation to myself.

    I mean, we’re talking on a physics web site hosted by a physicist who is pretty strange; just take a look at those claws. I feel confident that every crazy physicist (and there’s a lot on planet earth) will focus for at least a nanosecond on the very curios OPERA result.

    I can wait. I can’t wait; but I can wait.

  5. Could someone explain what the “start” pulse is for this experiment? I understood that it is a little transformer coil over the protom beamline. How is that calibrated?
    If there is -any- metal around this device, it could have a 60ns eddy current delay..
    It’s a pitty they haven’t trucked 1/3 of their detector back up to CERN to do the start experiment (and please use exactly the same cables marked for re-assembly so no lengths change. Hopefully that is done soon.

  6. Very interesting that they replicated the result with short pulses.

    Not a physicist so I probably misstate this, but I recall seeing a criticism after CERN’s first announcement that the use of satellite GPS system introduced a potential source of error in the timing. I think the concern was shifting frames of reference due to the rotation of the earth.

    Did this criticism get put to rest, maybe because GPS satellites use a geostationary orbit?

  7. Polkadot,

    You’d have to read through the paper to find it, but they explicitly state that although the previous GPS / atomic clock system that existed on-site had uncertainties associated with the measurements on the order of ~100 ns, the new system of synchronized atomic clocks has that beaten down to less than 3 ns.

    If they failed to account for general relativity properly, there may indeed be a systematic offset, but without releasing every specific detail of their calculations, only the insiders will know whether they did their work accurately or not.

    Since, as far as physics goes, that’s something that’s well-established and well-known, we take it as a given that they’ve done it correctly. It’s one of the things we’ll have to put our trust in: that they have competent physicists doing the “basic” physics work, where “basic” includes all of the effects of general relativity.

  8. How accurate does the distance measurement have to be in order precise results? Seems to me that a miscalculation of the distance of the ~730m of rock would lead errors. (Especially when dealing with 60 nano seconds)

  9. Now if they would just repeat this using the same measuring devices and detectors but with low-energy neutrinos, whose velocities are known with high precision from supernovas. If they see this 60ns offset with those neutrinos, it will prove there is a systematic error, even if they can’t find the error.

  10. Why don’t they use the same clocks and do the same experiment with light. If they find that light too appears 60 ns before it was supposed to, they would figure out that the problem is with the clocks or the distance measurement.

  11. Assume — just for a moment — that these measurements are accurate and correct. The implications of FTL neutrinos are even more mind-boggling than appears on the surface (the adjustment of the Special Theory of Relativity in a similar, if less significant, way to the “adjustments” made in turn upon Newton’s Theory of Gravity by the Special Theory).

    For instance, I read this over at Live Science (

    “String theory (has) broad-ranging implications, including the possibility that our universe has more dimensions than the known three dimensions of space and one of time.

    “String theory is incredibly difficult to test, and there is no proof that it’s correct. But if the neutrino measurements are correct, some physicists say string theory may offer the best bet of explaining them.

    “Perhaps, some physicists have suggested, the neutrinos are not traveling along the straight line we thought they were, but instead were hopping into one of the extra dimensions predicted by string theory, and taking a shortcut to their destination. If they traveled a shorter distance in the measured time, then their actual speed may not have been faster than light.”

    A possible proof of string theory, emerging from FTL neutrinos?

    First they have to prove that they ARE Faster Than Light….

  12. Thanks Ethan, the paper is surprisingly readable for a lay person. Their claim of 2ns timing accuracy sounds compelling given benchmarks with the 10 km tunnel, modeling of 1cm/year tectonic drift, and effects of the earthquake.

    Maybe we have to wait for independent replication.

    Keep up the good work.

    ps. seeing you in Wolverine costume reminds me of the 1995 movie “12 Monkeys”, a Hollywood sci-fi thriller starring Bruce Willis which (in my view) used M-Verse theory and it’s implication for time travel as the basis for a good yarn. Check it out sometime.

  13. Dear Ethan, thank you so much for mentioning my thoughts on this, even twice. Let me just add that what I say is not necessarily ‘speculations in case FTL are real’. The main point is that the data point towards the effect (FTL or systematic error!) being in the initial stretch (CERN, MINOS’s source) rather than having to do with the neutrinos traveling 730 km or timing issues.

    I do not believe that the 50ns binning at OPERA (the “jitter”) can be responsible for two reasons:
    1) The binning and statistical error should add up, or in other words, there would have to be a conspiracy between the neutrinos and the clock in order to give you the observed jitter.
    2) The energy dependence of the early arrival. OPERA’s previous results show 54 ns and 68 ns for 14 GeV and 41 GeV neutrinos.

  14. @12 Satish Ramakrishna

    Because that would involve drilling a 730km tunnel from Switzerland to Italy beneath the Alps. A project like that would make the whole CERN budget seem like pocket change.

    They aren’t shooting neutrinos trough some kind of tunnel. There is no way (yet) of creating a neutrino beam like we can make a photon beam (laser). In layman terms CERN creates a high energy proton beam which is shot into the wall, protons decay, all other particles get filtered by the rock, only neutrinos pass through. And then OPERA on the other side of the mountains detects them.

  15. Recently I performed a discrete 3D-matrix simulation of Maxwell-equations. I was astonished about the effect that with higher start centre-amplitude the propagating wave was much faster near its origin than further away. But I thought this effect belongs to problems of the discrete algorithm.
    With Sascha Vongher’s idea now I would suggest that something could be natural here, intrinsically hidden in the way energy-propagation actually happens in space-time.

  16. Robert S.: “It sounds like their clock system sucks if it has 25ns jitter, ”

    Agreed. For a 20MHz clock (i.e. 50ns period) a 25ns jitter is just plain horrible. But they probably use wrong terminology – they should call it “quantization noise”, not “jitter”. It should go down as the number of samples increases.

    Ethan: “Normally, you’d see a bell-shaped curve in the distribution of the arriving neutrinos. What we instead see is flat.”

    You can’t say much about the shape yet. Not with only 20 samples.

    P.S. My money is still on some trouble in modelling how the protons kick off the neutrinos…

  17. @Sinisa Lazarek (18):
    …just a Lat. D800 and only during week-ends. I used Excel-Makros to calculate and display – very slow. Calculation time was dependent on how far away from the centre I wanted to go – I used several dynamic algorithms, all calculating only for non-zero-cells – thus 100 steps needed about half an hour…

    I was inspired by this article and its related program “wqp.exe”…

  18. @17

    It is not rare to have wave that propagate faster than light. The problem is that some times you have to postulate an infinite wave to get the effect (if your wave is already everywhere, it can’t propagate), other times you have destructive interference at the borders, making them travel at the speed of light.

    Take a look at what wikipedia says.

  19. This would be balls-out awesome if it’s confirmed. As to the whole string theory alternate dimension angle…if Earth’s gravity is curving spacetime, and these neutrinos are spending all or part of their travel time in a dimension which is not curved, might that account for the 60ns discrepancy? By which I mean, a curved path ought to be longer, and can we calculate just how much longer this specific path in Earth’s field would be, and would this length difference be roughly 60 light-ns? Could be interesting….

    Not sure how this would jibe with supernova neutrinos. Most of their travel is through free space with presumably almost no gravitational field, but they certainly start off in an area with tremendous gravity.

  20. @Sinisa Lazarek – they don’t need to use visible light, just radio waves emitted from antennas stuck out of the ground. They really haven’t thought through ways to check their result here, real case of group-think.

  21. I’m not even close to being smart, let alone intelligent, but I’m smart enough to know this is going to be like hitting a hornet’s nest with a short stick.

    Scientists accept the measured speed of light, and accept it as a constant, and accept the principle that nothing can exceed it.

    Now the OPERA experiment suggests that neutrinos do, in fact, exceed the speed of light. That’s what’s referred to here in Texas as a “Rick Perry moment”: OOPS

    Louise Riofrio, at her blog GM=tc^3, maintains that the speed of light is slowing down. She has all the appearances of being knowledgable in her field.

    Could it be that she’s right? Has the speed of light decreased, but the speed of neutrinos hasn’t?

    Go ahead, beat me up – I deserve it.

  22. I was wondering something similar to the previous comment: is it possible that neutrinos move at the speed of light, but light that we have measured moves slower, as if the refractive index of a vacuum near earth is not 1. Then if light moves faster in interstellar space, then no one would have a good measurement of the speed of light, but neutrinos and photons from a supernova would still reach earth at about the same time.

  23. So if neutrinos turn out to be able to travel faster than light, does that mean they’re trying to be tachyons? Or just the frontrunners in a new trend. Get it tacky..ons…

  24. @Satish Ramakrishna – I’m qualified to comment on Political Science issues, but for radio physics its less so. But I would question the idea that timing signals that need to tick on a scale of nano-seconds could easily be transmitted 730km through the atmosphere in any readable fashion. If you could do something like that then I suspect you could massively simplify the infrastructure of the internet.

    And even if you could, the need to bounce it off of the ionosphere to reach your destination means your experiment would be a slave to atmospheric conditions. There is a reason why radio stations that are farther away have more static.

  25. @Satisha

    Agree with Thomas completely. Radio waves need a line of sight from emitter to receiver. Issue one: the earth is round, issue two: you have a 4.8km high wall between CERN and OPERA (the wall is Alps mountains). Don’t know of any 5km high antenna towers 🙂

    So the only thing is to bounce the signal from upper atmosphere… which will add so much noise and intereferance to the signal that it would be near useless.

    You say: “They really haven’t thought through ways to check their result here, real case of group-think.”

    Well.. perhaps because it doesn’t work that way. I would be careful to say something like that unless I had a equal or higher knowledge about the subject than them. In fact what they do use is GPS which would be your way but with added precision. So don’t understand how what would solve anything.

  26. When I was growing up in India, there was a whole network of microwave repeater towers that were being set up to connect the country electronically, for communication purposes – this was before the era of the satellite link. The advantage to the repeater network is that you can traverse large distances, don’t need bounces off the upper atmosphere and if the repeaters are built properly, you can control the cable delays – biggest advantage is that they are stationary on the earth, which I contend is the main problem with the clock synchronization (

  27. @30

    The only thing I could find that works similar to what you describe is this:

    But the accuracy is much worse than with GPS. Almost 200m as opposed to 20cm which OPERA says they have.

    But even if you could achieve (with radio) similar or better accuracy to that of GPS, that system would still suffer from possible systematics that OPERA (might) have. Mainly, you would still need to rely on geodesy measurements. How could you guarantee that the distance is precisely what you think it is? Also, and think this is more important you will have much more variables to the whole thing. Imagine 10 or so stations from CERN to opera. Some are at sea level, some are high in the mountains. Temperature, humidity, air pressure.. etc. Different small variations in equipment in all those different stations. Could you really account for all of them in all the circuits and wires in all 10 stations? Then on top of it all you have radio signals propagating through atmosphere. I personally believe that the final signal to noise would be several magnitudes higher than +-20ns.

    In theory at least, it might be possible. But doubt about the practical application. Think the “resolution” of radio waves isn’t high enough for these types of precise measurements. But that’s just my layman’s understanding. Am no expert on matter.

  28. Personally, think this is the last OPERA will do as far as experimentally trying different hardware setups. The ball is in the court of MINOS and others.

    It’s not like the guys from OPERA can go to budget people and say “Look… we really need a couple of billion euros more, to build some tunnels or outposts or whatever might be needed over pretty harsh terrain, ’cause we need a new clock.” 😀

    They are most likely gonna happily return to studying neutrino oscillations. Who knows, that might also shed some light on all this. 🙂

  29. Ethan Siegel wrote (November 18, 2011 1:55 PM):
    > […] Now, let me ask you this: if you’re a good scientist, what’s the next thing you’d demand?

    Put some appropriate instrumentation on the muon detectors in the CNGS beamline (shown on the right of “Fig. 2” above) — if it shouldn’t be in place already,
    figure in some suitable modelling of muons passing the antecendent hadron stop,
    and check whether muons were detected there as expected … (?).

  30. My understanding is that, regrettably, the time dilation experienced by light in Earth’s gravity well is not a big enough factor to discard for neutrinos and reconcile the results – that is, the idea that neutrinos aren’t slowed down by gravity isn’t enough. But I would appreciate if anyone knows where this calculation has been explicitly shown.

    Alternatively, an even wilder speculation: neutrinos actually have mass that is in some sense “negative” (but still unmeasurably tiny), and are sped up by a gravity well… hereafter known as the “double or quits” hypothesis.

  31. # 34: “Alternatively, an even wilder speculation: neutrinos actually have mass that is in some sense “negative” (but still unmeasurably tiny), and are sped up by a gravity well… hereafter known as the “double or quits” hypothesis.”

    Which would mean they hopped back in time, thanks to a gravity boost, and “arrived (at the other end) before they departed.” Mind boggling, but you learn to expect mind boggling things to happen in a quantum universe.

    I’m still hoping that the neutrinos took a short cut through embedded extra dimensions, thus going a good ways toward “proving” string theory, (insofar as proof is possible). But I’ve learned not to speculate when it comes to quantum mechanics. The impossible (like arriving before you leave) turns out too often to have been the case.

  32. I am not at all convinced of (a) the physical distance estimate and (b) that there is a systematic error in the timing; I was fulminating because the new experiments do nothing to address what I would consider the most obvious and most likely sources of error. The synchronization between the CERN source and the OPERA detector is by no means trivial. The internal timekeeping at OPERA is also a concern but quite frankly should not be as big a problem as the CERN-OPERA sync. What I would like to see is a different type of experiment. As for the 25ns jitter – that had me fuming too. I see it as absolutely intolerable; while GPS will have a part to play, the measurement accuracy of a second is better determined by a local reference clock; the jitter in the GPS signal should not be jerking the time about by 25ns.

  33. just a thought!

    is it just possible we are working in flat spacetime but spacetime is really curved – so the distance measurements are not accurate?
    Maybe neutrinos are so close to speed of light [from below:)] as to make this effect important for them but not for any heavier particles.

  34. @Satish Ramakrishna

    I’m aiming this comment at you because I think only you are in a position to appreciate it. I agree with your premise that the “common-view” synchronization technique is to blame. I also strongly agree with your statement in your arXiv paper that “if the two events were synchronized in the satellite frame (the clocks being set to the same time)…This means that the events are not simultaneous in the earth frame”.

    Please allow me to point you towards some information that may be very helpful to you:

    This document linked above is written by Giulia Brunetti who is one of the OPERA collaborators. You will find on pages 102 and 103 of the pdf document (or pages 94 and 95 as written on the document itself) a section titled “CERN-LNGS Intercalibration Measurements”

    In it Brunetti states that “…by taking into account the daily excursions of the GPS clocks, which for the Clock2 1PPS signal with respect to the Cs4000 reference could go up 60 ns (see figure 5.7).”

    This supports your statement that the Earth based clocks at CERN and LGNS are not in an inertial reference frame, and explains why a 60 ns excursion is found between them.

    Please email me at to continue this conversation, as I have further information that may be helpful to you.

    Many thanks,


    — 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.

  36. As I understand it Cern spent two years trying to find errors in the analysis before deciding to ask for help/replication. I think it’s unlikely the error is a very obvious one but I guess we’ll find out one day.

  37. The OPERA experiment likely has a timing error. Assuming there is no error, though, it is not likely that the detected particle travelled FTL, per se. That is, it is more likely that a tachyonic neutrino transmuted/oscillated along the way into an observable. Objections based on the “expected” energy spectrum (cf. ICARUS scientists) do not take this into account. Such objections are not only inconclusive, they beg the question.

  38. You’re doing this wrong.

    Think about how it COULD be going faster than light.

    Work out what you’d have to do to prove or disprove this theory.

    Make an experimental test.

    Check to see if you’ve proved, disproved, or left ambiguous the theory.

    The advance of science comes after an experimental scientist says “that’s odd…”. It stops when they say “that cannot be right”.

  39. What if the speed of light isn’t constant?

    What if the speed of light varies through time and space?

    That would create some interesting theory. At least I think so.

    Antimatter is the mind and consciousness of all living entities.

    You are your own universe.

    Reality is where the minds (antimatter) meets the physical universe.

    Interested? Then read my philosophical multiverse theory.

    Google crestroyer theory, and find it instantly.

  40. Is it possible that what has been considered the “speed of light in a vacuum” is not actually the actual speed of light in an an actual vacuum because space is not actually a vacuum considering matter, dark matter and dark energy? Perhaps the true speed of light is slightly faster than our measurements.

  41. Why egg on their faces? OK – so if demonstrated wrong, they must have made some error somewhere. The result of their research is/was unlikely. But does that mean it shouldn’t have been published? Never admit to an outlier? So why the ‘Ya boo! There’s egg on your face!’ schoolyard chant? Why are people not allowed to make an honest error without losing face? I never could understand the ‘cold fusion’ disgrace effect. Nothing to do with science, only with egos. A good scientist should be above that kind of stuff.

Comments are closed.