A Fireball Drifted Down the Aisle of His Plane

What We Know

At five minutes past midnight on March 19, 1963, lightning struck the Eastern Air Lines flight R. C. Jennison was riding from New York to Washington, and the cabin lit up in a sudden, loud flash. What came through the flight-deck door next held its shape the entire length of the aisle: a sphere about 22 centimeters across, holding a steady course, passing some 50 centimeters from his face before it merged into the skin of the fuselage near the tail. Jennison was a radio astronomer at Jodrell Bank, not a nervous flier prone to seeing things, and he put numbers on what he’d watched: roughly 1.5 meters a second, glowing at 5 to 10 watts. He published the account under his own name, in Nature.

The phenomenon already had a track record by 1963. It got its name, “foudre globulaire,” in 1837, when the French physicist François Arago gathered enough independent accounts of glowing, free-floating spheres during thunderstorms to treat them as one real thing worth studying, rather than dismiss them as folklore. For most of the two centuries since, that’s what the evidence has consisted of: descriptions. Vivid, specific, often from credible witnesses. Still descriptions.

That changed, partially, on July 23, 2012. A team led by physicist Jianyong Cen had set up video cameras fitted with diffraction-grating spectrographs on China’s Qinghai Plateau, hoping to catch an ordinary lightning spectrum. Instead, for 1.64 seconds, they caught something else: a ball of light drifting away from a strike, its glow broken into wavelengths, something no natural sighting had ever offered before. The spectrum showed iron, calcium, silicon, nitrogen, and oxygen: the signature of vaporized soil. It was the first natural ball lightning event ever measured by an instrument built to say what it actually was, rather than by a witness’s memory of what it looked like.

Two data points, 175 years apart, an ocean and half a world between them, describing what sounds like the same thing. Nobody disputes that something real is happening. What it is stays open.

The Roadblock

Ball lightning is rare, unpredictable, and usually over before anyone can react: the Qinghai event lasted under two seconds, and most witness reports put the whole encounter under a minute. It also arrives the way a thunderstorm does: without warning, at a location nobody can pre-position a spectrograph for. In nearly 190 years of a named scientific problem, exactly one natural event has ever been captured by an instrument capable of measuring what it was made of. Every other data point in the historical record is somebody’s recollection, however careful.

That gap has pushed the real work into laboratories, and labs have obliged: electric arcs, lasers, microwaves, burning silicon have all produced short-lived glowing spheres that resemble ball lightning to varying degrees. The trouble is that resembling it and being the same physical process are different claims, and no lab reproduction has ever had its spectrum compared directly against the one natural event anyone actually measured. Researchers aren’t just arguing about the cause anymore. They’re arguing about whether their own machines are studying the same phenomenon at all, or four separate curiosities that happen to glow and float.

Best Guesses

Silicon-vapor combustion

This is a ground-strike mechanism by definition, which leaves Jennison’s midair sphere unexplained on its own terms — but it has the best physical evidence of any theory here. When lightning strikes silicon-rich soil, the heat vaporizes silica; carbon in the dirt strips away the oxygen, leaving loose silicon nanoparticles that drift upward and slowly re-oxidize, releasing the stored energy as heat and light for several seconds. The 2012 Qinghai spectrum matches the predicted elements almost exactly — the strongest physical evidence any of these four theories has behind it.

Trapped-microwave soliton

A lightning stroke’s leading edge produces a burst of relativistic electrons. When that burst strikes a surface, the theory goes, the resulting radiation can ionize a spherical pocket of air and become trapped inside it, sustaining a glowing bubble with no need for ground contact. It’s the only one of these four built from the start to explain a sighting like Jennison’s, inside a sealed aircraft. In 2026, a laser lab in Shanghai produced a structure with roughly the right shape and spectral fingerprint: 400 microns wide, a peer-reviewed first. It lasted 100 nanoseconds. Whether that scales to a sphere loitering in a cabin for several seconds remains an argument, not a result.

Electric-discharge plasmoid

Since 2004, researchers running high-voltage discharges through an electrolyte have generated thousands of floating, glowing plasma spheres, 12 to 20 centimeters across — one of the most reproducible ball lightning effects ever made, on demand, in a lab. They fade in about a second, though, and need a liquid conductor no thunderstorm sky has ever supplied. Whether the resemblance holds up despite that gap is now a live argument: a 2025 exchange in an atmospheric-physics journal disputes it, and neither side has backed down.

Optical or sensor artifact

Not every sighting is a sighting. In 2025, researchers at Texas State University ran computer-vision analysis on security-camera footage of two “ball lightning” objects filmed during a 2023 Montana storm, and found that a mundane origin could not be ruled out: burning debris from a nearby power-line arc, its size inflated by the camera sensor’s own glare. Viral footage gets undercut on similar grounds on a regular basis. It’s a genuine, useful corrective. It also has nothing to say about a trained physicist, fifty centimeters away, with no camera between him and what crossed his cabin.

The Verdict

The one instrumented capture the field has ever gotten was an accident, a spectrograph aimed at something else that happened to catch this too. Nobody has built the alternative on purpose: a spectrograph paired with a fast camera, deployed widely enough that the next multi-second sphere doesn’t just get remembered — it gets measured, for composition, duration, whether it needs soil nearby to exist at all.

That’s an engineering problem, not an unsolvable one, and it would do more for this question in a single lucky capture than two centuries of testimony has managed. Until someone builds that network and gets lucky twice, the field stays stuck weighing eyewitness memory against laboratory curiosities that only sometimes resemble each other, arguing over which resemblance counts. Jennison’s sphere crossed his cabin in well under a minute, and no instrument was there to catch it doing so. Most of them still aren’t.