Friday, September 30, 2011

Nir Shaviv: Why don't I believe that neutrinos travel faster than the speed of light?

Guest blog by Prof Nir Shaviv (HUJI), astrophysicist

Reprinted from with author's permission

Three weeks ago I visited the underground laboratory of Gran Sasso near l'Aquila. Little did I know that it would make headline news so soon, for "discovering" particles moving faster than the speed of light. Since a few people asked me what did I think of it, I decided to write something about it here.

Before I get into the details, here's a recap of the news item.

The OPERA experiment, which is one of a dozen or so running at the underground laboratory at Gran Sasso in Italy (the largest underground laboratory), came out with sensational results: they measured neutrinos traveling faster than the speed of light. Now the details.

The OPERA experiment was built to detect 17 GeV or so neutrinos which are produced at CERN, and aimed towards Gran Sasso 700 km's away. The idea is to try and find muon neutrinos that "oscillated" and changed into tau neutrinos. This can be expected because neutrinos appear to have a small mass. (For the physicists of you, this is a consequence of the fact that the "flavor" eigenstates are not the mass eigenstates – this is akin to two pendula coupled with a very week spring, as a consequence, if one perturbs one pendulum, eventually the oscillation passes to the other pendulum and back). Anyway, they have thus far detected one(!) such event (see

Besides measuring the neutrinos and their flavor (using strong magnets, to see how massive is the particle that they form [whether it is a muon or a tau particle, both of which are heavier brothers to the electron]), they also measure the timing of the events. This they can compare to the expected time it takes the neutrinos to travel between CERN at Geneva and Gran Sasso near l'Aquila. Under standard theory of special relativity, because the neutrinos are extremely relativistic, this time should be very nearly the time it takes light to cross this distance. However, they found that it took the neutrinos less time to cover it, 60 nanoseconds less (which is equivalent to cover 18 meters more). This implies, if the measurement is right, that the neutrinos have travelled faster than the speed of light. Is this correct?

The gut feeling I have when I hear something like that, is that there is something wrong in the experiment. The reason is that the speed of light is so deeply engrained in physics, that giving it up would require very convincing results. This result, on the other hand, is not. And why is it so?

To begin with, a good experiment to measure anything which is not "zero" (e.g., the difference of something from the speed speed of light, a non-vanishing mass) is to plan an experiment that gives a signal if the measured quantity does not vanish, and no signal if there is no difference. One classic example is the Michelson-Morley experiment. The set up was such that there would be no change in the diffraction fringes if there is no compressed aether, and a changing pattern if there is (all this, without having to accurately know the length of the arms in the experiment or their difference).

The present experiment simply measures a very large number, which is the time it takes the neutrinos to travel from CERN to Gran Sasso, and then this large number should be compared with another large number, which is the time it takes light to travel this distance. One should then hope that both are calculated/measured accurately enough such that their difference is calculated with a high enough accuracy.

The second reason why my guts tell me that the measurement is not due to superluminal neutrinos is because (anti-)neutrinos at a lower energy (of several MeVs) were measured to have a speed which is very nearly that of the speed of light, at one part in a billion. This means that this superluminal effect so happens to kick in at the right energy for OPERA.

The third reason is that particles at this energy ballpark were already extensively measured. It would have been a different story if there would have been a claim that cosmic rays at 1021eV are superluminal, because that is a regime which wasn't studied on Earth.

Bottom line, I find it hard to fathom that neutrinos are superluminal. But I don't have a good explanation for the measurements either. (Frankly, I have other things to think about.) So, what can the source be? Look at the list of possibilities summarized in Luboš Motl's blog, perhaps one of them is the reason.

The author standing in front of the best OPERA in town, 1400 m below the surface, in the Gran Sasso Laboratory. Click to zoom in.

Let me end with the first super-luminal neutrino joke I have heard:
"We don't allow faster than light neutrinos in here," said the bartender. A neutrino walks into a bar.

LM: You may also listen to a fresh radio interview with Lisa Randall, on neutrinos, Tevatron, LHC, and her book, on Ira Flatow's NPR radio show (plus transcript).

LM: You may also check a paper by my fellow birthdaymate Sheldon Glashow and Andrew Cohen, an ex-co-author of mine, who argue that superluminal neutrinos are impossible also because they would emit eine Bremsstrahlung.

This is a cute argument but I guess that those who are eager to abandon relativity will have no problem to claim that the formulae to quantify the intensity of der Bremsstrahlung near the speed of light should be modified, too. After all, the breakdown of relativity would lead to many other pathologies. For example, one could construct perpetuum mobile devices employing black holes which would have different horizons for different particle species (neutrinos vs photons etc.).

LM: In another preprint, Carlo Contaldi elaborated upon one of the mistakes suggested by your humble correspondent (search for "simultaneity"), namely that the experimenters could have been sloppy about the definition of "simultaneity of two events" which is important for synchronization and which, according to special relativity, depends on the speed of your inertial system.

My numbers indicated that they would have to incoherently mix the static Earth's frame with the satellites' frame (i.e. to synchronize the time from the satellites' viewpoint while using the Earth's frame in other contexts) – the speed of the satellites is enough to produce 30 nanoseconds in this way (one half of the OPERA speed excess). However, in my opinion, the speed of the Earth's rotation itself (and the time difference produced by this speed out of 730 km via the Lorentz transformation) isn't enough to explain away most of the 60 nanoseconds: the speed is just too low.

Contaldi seems to disagree but one probably has to study his paper in more detail to see what is the exact effect he claims to have been neglected.


  1. The Glashow / Cohen paper is hilarious in that part of its argument involves imagining a frame of reference moving essentially at the speed of light. What morons.

    I sure hope that the effect stands. In fact I wrote as much 4 years ago in a guest post at Tommaso's blog: ""Every now and then I get the urge to include a first slide at a physics conference that says “Einstine was Wrong”, preferably in a green crayon. So far I’ve suppressed it. I think that eventually we will find plane waves that travel faster than c, and that the evidence will be undeniable. Until then, I really don’t think that there is any way anyone can be convinced by these sorts of arguments. On the other hand, I didn’t think my guest post would convince anyone either, LOL." Let's see if your blog will let me link in the prescient comment:

  2. "What morons." - This is a little bit exaggerated, Carl, isn't it? According to everything I know, you can think about physics from the viewpoint of *any* reference frame connected with any velocity different than "c", however high. Of course, this assumption invalidates the big OPERA claim immediately.

    "Einstine" - the common crackpot spelling of Einstein's name is Einstien, Carl. You're thinking outside the box! ;-)

  3. “Einstine was Wrong” But Einstein was right lol

  4. This haiku's title is my own, originally coined word.

    -- James Ph. Kotsybar

    Sometimes particles
    can go faster than light speed
    and backwards in time.

  5. LM, can you please explain in simple terms why a superluminal neutrino at 17GeV would shed energy rapidly as Cohen & Glashow compute; but a subluminal neutrino at the very same energy would not?

    I guess the argument has to do with moving a charge (any of SU(2) x U(1)) faster than the speed of light and something Cerenkov-like, but it is not clear whether I have to assume something about the kinematics of a superluminal particle or not. I would prefer not, because then the force of the Cohen-Glashow argument is so much more.

  6. "Of course, this assumption invalidates the big OPERA claim immediately."

    Hey, that was MY point! By the way, the internet is alive with Einstien spellings. Pick your favorite school and I bet you'll find them there too. It's almost a standard spelling so I chose a more extreme version.

  7. I posted on arXiv a possibly very simple explanation of OPERA results without any strange physics or measurement error. It is enough that the probability detection changes during the leading/trailing edges, for instance by a change in energy composition of the beam, to produce a shift of the order of the rise/decay time. The measured effect is only 10 % of this rise time, so it is perfectly conceivable.

  8. I posted on arXiv a possible very simple explanation of OPERA results without any strange physics or measurement error. It is enough that the probability detection changes during the leading/trailing edges to induce a shift in the fitted time. A change of the beam energy distribution during these phases could easily produce such an effect. Contrary to the velocity shift, this effect would depend on the edge shape, but not on the travelled distance.

  9. "The Glashow / Cohen paper is hilarious in that part of its argument involves imagining a frame of reference moving essentially at the speed of light. What morons."

    No, it doesn't. The computation of both the pair-emission and energy-loss rates is done using SM amplitudes. That computation gives, in particular, the numerical constants k, k'.

    The other, dimensionful factors can be obtained by dimensional analysis plus simple physical arguments. But that's just a side comment, not the main point of the argument.

    By the way, I think Glashow has a card or two proving he's not a moron. I don't know anything about his co-author.

  10. Dear Ambrose, I know Sheldon Glashow from dozens of physicists' fancy dinners etc. and I can confirm he is not a moron.

    Andrew Cohen is not a moron, either, as I can confirm as a co-author of a physics paper. ;-)

  11. I must admit the Glashow/Cohen paper strikes me as a bit dismissive of the obviously impressive effort made by the Opera team. They simply state formulas without any indication what kinematical rules they're assuming. E.g., they say "the 4-momentum of the superluminal neutrino is timelike (relative to the speed of light)" - a statement I could not parse at all.

    I didn't study the Opera paper for some time because I assumed there was some complicated or silly relativity error in it or the GPS system, but I did not realize that the GPS system is set up quite simply in the earth center-of-mass frame, sidestepping any real complications from relativity. But I do wish they had double-checked it with a physically transported clock, instead of the "even better" GPS-based system that the metrology agency used.

    As you say, only the satellite speeds are high enough to affect the result, but they don't seem to be relevant given how GPS is set up.

    Well anyway, I think the Glashow paper could have been written a bit more respectfully. That's what I'm really grouching about. The experiment seems like a tour-de-force even assuming that some error is ultimately discovered.