OPERA 2011 — when a loose cable beat Einstein for six months

The setup that produced the 2011 saga was, on paper, an exemplary use of GPS-disciplined timing. The CERN SPS extracted 400 GeV protons in two 10.5-microsecond spills six seconds apart; the protons hit the carbon target of the CNGS beam line; the secondary pions and kaons decayed in flight to produce muon neutrinos pointed downward into the crust at 3.2° below horizontal; the beam emerged 730 kilometres later at the Gran Sasso underground laboratory, where the OPERA spectrometer recorded the interactions that survived the trip. The geometric distance, surveyed by the CERN geodetic group through three independent measurements over the crust and through GPS triangulation at both endpoints, was known to about 20 centimetres. The time at which protons left CERN was tagged by a dedicated time-of-flight measurement on the SPS extraction waveforms, synchronised to UTC through GPS receivers at CERN and Gran Sasso connected to a common-view algorithm that cancels most ionospheric-delay systematics. The expected travel time for light over 730 km is 2.4 ms; the additional rock path adds nothing because light through rock is just slower light through air, irrelevant to a vacuum-c comparison; the neutrinos, if they travel at exactly c, should arrive at the master clock simultaneously with a hypothetical photon shot through the same path.

OPERA's collaboration claimed in the autumn of 2011 that across 16,111 events accumulated over three years of running, the neutrinos arrived 60.7 ± 6.9 (statistical) ± 7.4 (systematic) nanoseconds earlier than the c-prediction. The corresponding apparent deviation from the speed of light was (v_ν − c)/c = (2.48 ± 0.28 ± 0.30) × 10⁻⁵. The result was announced at a CERN seminar in September, posted on the arXiv the same day, and made the front pages of essentially every newspaper on Earth within forty-eight hours.

What the supernova said

The first reaction in the community was that the result was probably wrong, and the most powerful argument was almost a quarter-century old. SN 1987A, the supernova in the Large Magellanic Cloud whose neutrino burst was detected at Kamiokande-II and IMB on 23 February 1987, had constrained the neutrino speed to equal c at the level of about one part in a billion. The neutrinos and the photons from the explosion arrived within roughly three hours of each other after a 168,000-light-year journey. If neutrinos at MeV energies had travelled even ten parts per million faster than light, they would have arrived years before the photons rather than hours; the SN 1987A observation alone constrained the speed deviation to below 10⁻⁹ for MeV neutrinos. OPERA's claim was about 10 GeV neutrinos rather than 10 MeV neutrinos, and energy-dependent speed-of-light violations could in principle reconcile the two, but every plausible theoretical extension that allowed this introduced other observational problems. The community settled into a careful, semi-public hunt for the systematic error.

The interactive below shows the geometry of the timing problem. The travel time for light over 730 km is fixed at 2.44 ms. A hypothetical timing offset in the kit — caused by a slow GPS receiver, a stale fibre-optic delay calibration, or a clock that is recording arrival times late — translates linearly into an apparent superluminal signal. Slide the offset to see the resulting (v − c)/c, and compare to the constraints SN 1987A and IceCube already had on file.

The slider's lesson is unsubtle. The leverage from the long baseline is enormous: a single nanosecond of timing error translates into a (v − c)/c of about 4 × 10⁻⁷, well above the SN 1987A bound on MeV neutrinos and within reach of long-baseline neutrino experiments at GeV. A sixty-nanosecond offset — well below the few-microsecond GPS jitter that everybody worries about and well above the dedicated calibration error budget — produces the entire OPERA 2011 result. That is what made the search for the systematic so difficult: any single piece of the timing chain that was off by a fraction of the experiment's overall calibration accuracy could fake the effect.

The cable

In February 2012 the collaboration announced that a fibre-optic cable connecting the external GPS antenna receiver to the master clock card on the OPERA timing rack had been found to be incompletely seated in its connector. The resulting signal delay had been characterised at about 73 ns, in the same direction as the apparent superluminal signal. When the cable was correctly seated and the data reprocessed with the updated calibration, the apparent superluminal advance dropped to within errors of zero: the corrected result, published later in 2012, gave (v − c)/c = (2.7 ± 3.1 stat ± 3.4 syst) × 10⁻⁶, fully consistent with light speed. The independent measurement by the ICARUS collaboration at Gran Sasso, working from the same CNGS beam in the same period, was consistent with this corrected number and inconsistent with the original claim. The OPERA collaboration's spokesperson and physics coordinator resigned over the affair. The detector itself went on to its primary science result — direct observation of ν_μ → ν_τ appearance, eventually published as the 6.1σ detection of ten kink events between 2010 and 2018.

What the saga is and is not a lesson about

The most-quoted lesson — "always check your cables" — undersells what happened. OPERA's 2011 result was not produced by a single careless person. It came from a six-month dedicated effort to nail down systematics that the collaboration had every reason to take seriously, with a documented error budget at the few-tens-of-nanoseconds level. The cabling problem was below that budget by a factor of about two and was identified only when external scrutiny forced an end-to-end re-examination of every link in the timing chain. The episode is sometimes invoked as evidence that the announcement was premature; the collaboration's own response was that the data had been examined, the systematics had been quoted honestly, and the appropriate next step was to put the result in front of the community so that the inevitable detailed cross-checks — by them and others — could happen. That defence has merit. The error was real, but the response to it — public, fast, with the responsible parties stepping down rather than litigating — is one of the cleaner examples of how scientific self-correction can work when nothing is being defended for its own sake.

For the actual physics OPERA went on to do, see OPERA — catching ν_μ turn into ν_τ in flight; for the SN 1987A measurement that set the prior nobody could shake, see SN 1987A — the 13 seconds that founded neutrino astronomy.