Earthquake Lights: The Glow in the Sky That Science Refused to Believe
On the evening of November 12, 1988, people living along the St. Lawrence River near Quebec City looked up and saw a bright globe of pink-purple light crossing the sky. It was low, and it was silent. It did not flash and disappear the way lightning does. Witnesses described a steady, deliberate drift toward the north-northeast, following the line of the river, lasting two or three full minutes before the light simply went out.
Thirteen days later, at 6:46 on the evening of November 25, a magnitude-5.9 earthquake struck the Saguenay region roughly 150 kilometres to the north. It was the strongest tremor eastern Canada had felt in decades and was registered across much of the continent. A magnitude-4.8 foreshock had arrived two days earlier, in the small hours of November 23. Between November 1988 and the end of January 1989 the region produced sixty-seven separate earthquakes.
The Saguenay shock cracked buildings, buckled roads and set off landslides, and it became a reference event for seismic hazard in eastern Canada. Its size is also why the sky reports survived. A large earthquake that is studied closely draws attention to everything that surrounded it, and in this case that included what people had seen overhead in the days before and the weeks after.
A Quebec researcher, France St-Laurent, set out to collect those sightings. She eventually assembled forty-six reports detailed enough to analyse and published them in the journal Seismological Research Letters. The catalogue she built is unusually specific. Witnesses described balls of light floating about a metre above the ground. They described motionless shapes that looked like meteors, some trailing streamers, appearing to emerge from the earth itself. They described rays and broad bands standing across the sky. St-Laurent measured each account against a classification scheme drawn up in 1948 by the Swiss scholar Frederick Montandon, and against a description published by a Japanese researcher in 1968, and found that the Saguenay sightings fell into categories defined long before anyone in Quebec saw anything.
Across the documented record the timing varies widely. Some lights appear during the shaking itself, some in the minutes or hours before it, and a small number, like the Quebec globe, days ahead.
Those older frameworks existed because the reports are old. Long before earthquakes had a science, chroniclers in Europe, Asia and the Americas recorded the sky behaving strangely while the ground shook: bluish flames rising out of the earth, glowing orbs hovering over valleys, sheets of pale light standing on a cloudless horizon. Cultures with no contact with one another produced descriptions that overlap closely. Montandon's 1948 work, which sorted these accounts into formal categories of what he called geo-atmospheric luminous phenomena, was an early attempt to treat centuries of folklore as data.
The first serious body of modern evidence came from Japan. When the Izu Peninsula earthquake struck in the dark before dawn on November 26, 1930, killing hundreds, people across a wide swath of the country reported the sky and the ground lighting up: beams standing on the horizon, fireballs, broad sheets of pale glow, streaks that moved. The physicist Torahiko Terada took the accounts seriously enough to publish a formal study of them in 1931, proposing a mechanism he called streaming potential, in which water forced through rock under enormous pressure carries electric charge to the surface. His contemporary Kinkichi Musya went further and painstakingly gathered roughly 1,500 separate reports of luminous phenomena tied to that single disaster. His conclusion was blunt. The observations, he wrote, were "so abundant and so carefully made that we can no longer feel much doubt as to the reality of the phenomena."
Testimony alone did not settle the question. Photographs did more. During the Matsushiro earthquake swarm, which shook central Japan with hundreds of thousands of tremors between 1965 and 1967, the luminous phenomena were finally captured on film. Yutaka Yasui of the Kakioka Magnetic Observatory analysed the images and described them in a landmark study published in 1968, the same description St-Laurent would later use as a yardstick in Quebec. In 1973 John Derr, a seismologist with the United States Geological Survey, reviewed the accumulated record and judged the phenomenon well established.
Video eventually did what photography could not. When a magnitude-8 earthquake struck near Pisco, Peru, in August 2007, security cameras in Lima, 150 kilometres from the epicentre, recorded flashes lighting the night sky as the ground rolled. A team at the Pontifical Catholic University of Peru led by Jorge Heraud compared the footage frame by frame with seismic records and showed the light pulsing in step with the passing seismic waves. For the first time the timing of the glow was tied to an instrument trace rather than to human recollection.
Other cases followed in quick succession. In L'Aquila, Italy, residents reported small flame-like flickers above a paved street in the hours before the deadly earthquake of April 2009. In September 2021, thousands of people in Mexico City filmed blue and green bursts flaring overhead as a magnitude-7 earthquake struck near Acapulco. In September 2023, minutes before a magnitude-6.8 earthquake killed nearly three thousand people in Morocco, cameras recorded blue flashes crossing the night sky, and the footage travelled around the world before the aftershocks had faded.
In 2014 the subject received its most substantial scientific treatment to date. Writing in Seismological Research Letters, Robert Theriault, France St-Laurent, Friedemann Freund and John Derr sifted centuries of accounts and retained sixty-five cases since 1600 that they judged solidly documented. The geographic pattern surprised them. Roughly 85 percent had occurred in continental rift environments, and about 97 percent along near-vertical faults, including the faults that transect the Saguenay Graben, the ancient rift valley beneath the 1988 sighting.
Laboratory work has established one piece of the physics beyond dispute. Rock under great stress does generate measurable electric current. In 2010 a Japanese team photographed faint flashes of light escaping from a block of granite at the instant it was fractured, a small and fully controlled example of rock that glows when it breaks. Freund, a mineral physicist whose career passed through NASA's Ames Research Center, the SETI Institute and San Jose State University, argued that the same chemistry scales up: under stress, defects in the rock called peroxy bonds break apart and release positive charge carriers, which travel to the surface, build intense local electric fields and ionize the air above. Ionized air glows. The laboratory current is documented. The step from a fractured granite block to a lit sky above a river valley is not.
Conclusions and Open Questions
The documented record is not in serious doubt. What the lights are remains contested, and each leading explanation carries a specific weakness.
Freund's peroxy-bond theory is the most developed. It accounts for the geography reported in the 2014 catalogue, since deep near-vertical faults in rift rock would be efficient at delivering charge to the surface, and it rests on a real laboratory effect. Its weakness is that no instrument has ever measured a surface electric field strong enough to ionize open air above a fault, and nothing in the model explains a glow with thirteen days of lead time. Terada's older streaming-potential idea has the opposite problem: it requires a specific arrangement of water and fractured rock that is hard to confirm underground. Piezoelectric charge from quartz-bearing rock is genuine in the laboratory but notoriously difficult to scale to a hillside. Radon seeping from stressed ground can ionize air, but measured emission rates appear far too low to paint a sky. Static discharge from fracturing rock produces light on the wrong scale.
Against all of these stands the skeptical case, and it is strong. Many geophysicists argue that most modern footage records ordinary electrical infrastructure. Power lines arc, transformers explode, and low cloud turns blue at precisely the moment shaking begins. The United States Geological Survey remains deliberately neutral, noting that geophysicists still disagree over how many reports describe a genuine phenomenon at all. The lights filmed during the 2023 Turkey-Syria earthquake resolved on inspection into a mix of an ordinary thunderstorm, electrical discharges and fires. Most experts explain much of the Mexico City footage the same way, and many suspect the Morocco flashes were the same. The weakness of this explanation is equally clear. It cannot touch Quebec in 1988, where the grid was intact and nothing had yet moved underground, and it cannot touch the Matsushiro photographs taken in rural Japan.
A further problem undermines the whole catalogue. A glow seen before an earthquake can always be linked to it afterwards. Nobody records the identical pink globe that drifts over a valley which never shakes. The historical catalogue is therefore built from the survivors of a selection process, and no study has yet corrected for it.
There is also a geographic contradiction that is rarely stated plainly. The 2014 catalogue points to continental rifts, while almost every famous video comes from a subduction coast: Peru, Mexico, Morocco. One theory holds that this is not an error in either dataset but evidence that two different things are being recorded, a rare geophysical glow tied to deep faults and a far commoner urban light show that only resembles it on a phone screen. Some argue the implication is broader still, that earthquake light was never one phenomenon and that a century spent hunting a single mechanism was a category error.
The open questions are concrete. What process can produce light days before a rupture? Why has no field instrument recorded the electric field the theory requires? How many unremarked sightings never entered the record at all? And if the lights are genuine precursors even part of the time, they would amount to a natural alarm, which is exactly why the field is careful about saying so. Cameras now cover the planet. Either unambiguous footage eventually arrives from a dark valley far from any power grid, filmed as the fault beneath it fails, or it never does, and that silence will be an answer of its own.