mistake.club2nd Opinion ↗
Back to the archive

The encyclopedia · R&D & Science · Strategic decision · 2015–2016

CERN's 750 GeV bump: 500 papers about a statistical wiggle

Both LHC experiments saw a bump in the same place; five hundred theory papers later, more data showed it was never there.

CERN · 2015-12-15

What happened

On 15 December 2015 CERN held its showpiece seminar for the first results of the LHC's second run. ATLAS and CMS, the two big collider experiments, both presented a small excess of photon pairs at a combined mass of 750 GeV — a bump in the same place, in two independent detectors. The local significances, around 3.9 sigma for ATLAS and 3.4 for CMS, were a long way from the five sigma particle physics demands for a discovery.

The reaction was immediate and global. 'If this thing turns out to be real, it's a ten on the Richter scale of particle physics,' said John Ellis, former head of theory at CERN. Theorists churned out more than 200 papers within three months and around 500 by August, proposing gravitons, axions, composite Higgs particles and more. Physical Review Letters' editor Robert Garisto published an editorial about the onrush and decided to print just four of the explanations.

The corrections arrived before the verdict. At a conference in March 2016 the collaborations updated their analyses: CMS's bump fell from 3.4 to 1.6 sigma after technical corrections, ATLAS's to about 2 sigma. 'I'm afraid the most likely resolution is that what's been seen is a statistical fluke,' said Nobel laureate Frank Wilczek. CMS head Tiziano Camporesi offered colleagues 20:1 against the particle's existence, in bottles of decent French wine — 'I am betting against it, but I would gladly lose.'

On 4 August 2016, at the ICHEP conference in Chicago, the collaborations presented the 2016 data — nearly four times more than 2015's. The bump was gone. 'We see many events but we see no bump,' said ATLAS spokesperson David Charlton; ATLAS put the compatibility of the two years at 2.7 sigma. CMS member Tommaso Dorigo's verdict: 'theorists should have known better.' Five hundred papers had been written about a fluctuation that more data erased in a day.

Why it happened

  • The hint was staged as an event: first Run 2 results, a showpiece seminar, two experiments with the same bump — a 3-sigma anomaly read as a near-discovery before the corrections.
  • Five sigma exists for a reason: at three to four sigma local, and less after the look-elsewhere effect, a statistical fluke is the most likely explanation, not a new particle.
  • The theory machine ran on the hint with nothing new to test: hundreds of papers bloomed between December and August because the only data that could settle it did not exist yet.
  • The March corrections were public and fatal — CMS fell to 1.6 sigma — yet the story survived until August, because no experiment had said the word 'fluctuation' on the record.
What it cost500 theory papers and a year of the field's attentioncostly

The lesson

A bump is not a particle. Five sigma is not bureaucracy: it is the line between a discovery and a story, and the LHC has seen many three-sigma stories come and go.

Aftermath

Physical Review Letters printed four of the hundreds of explanations, with an editorial about the onrush; Camporesi collected no wine; and the field went back to work — the bump's real legacy was a running joke about theorists who 'should have known better' and a sharper sense of what the five-sigma bar is for. The episode also underlined how much the community wanted the LHC to find something new: the desire itself, Wikipedia noted, was part of what made the bump so easy to believe.

Sources

spotted an error? The club wants to know.

Comments · 0

    Sign in to join the comments.

    More like this

    2011–2012 · R&D & Science

    Neutrinos beat light by 60 nanoseconds — the fault was a loose cable

    An experiment's credibility and 'Einstein wrong' for months
    2023 · R&D & Science

    LK-99: the 'world-first' superconductor that was copper sulfide and iron

    24 years of work, three weeks of hype, one credibility bill
    2014–2015 · R&D & Science

    The smoking gun of cosmic inflation was a smudge of galactic dust

    Ten months of Nobel talk for a galactic dust signal

    Somewhere, someone solved the problem this company failed at. 2nd Opinion →