In July 2012, researchers from Northwest Normal University in Lanzhou, China, were filming ordinary cloud-to-ground lightning on the Tibetan Plateau when they accidentally captured something that had eluded scientific instruments for centuries. A glowing sphere, roughly five meters wide, appeared at the point where a bolt struck the ground, drifted horizontally for about 1.64 seconds, and faded. They got it on digital video. They got its optical spectrum. The spectrum showed silicon, iron, and calcium—elements found in soil, not in the atmosphere—which was the first empirical evidence from a natural occurrence supporting a specific hypothesis about what ball lightning actually is.
That was 2012. It remains, as of 2026, the only scientifically instrumented recording of what is widely believed to be genuine ball lightning. One data point. From one event. Lasting less than two seconds. For a phenomenon that has been reported by thousands of eyewitnesses across centuries, on every continent, in conditions ranging from open fields during thunderstorms to the interior of sealed aircraft at cruising altitude.
What the witnesses describe
The “average” ball lightning—if an average can be constructed from thousands of anecdotal reports and almost zero instrumented data—appears as a luminous sphere roughly 10 to 30 centimeters in diameter, glowing with the brightness of a 100-watt lamp, lasting about 10 seconds. It typically appears during or immediately after a thunderstorm, often near the point where conventional lightning strikes the ground. It moves parallel to the earth’s surface, sometimes slowly drifting, sometimes bouncing, sometimes hovering motionless. It can move with the wind, against the wind, or in no discernible relation to the wind at all.
The properties that make ball lightning genuinely strange—and genuinely difficult to explain—go beyond “glowing sphere in a thunderstorm.” Witnesses report that it passes through closed glass windows, sometimes leaving small holes roughly a third of the time and sometimes leaving the glass completely intact. It has been observed indoors, appearing to materialize in enclosed rooms with no obvious entry point. It has been reported inside sealed aircraft. It can change shape, compressing through openings much smaller than its diameter and reforming on the other side, behaving less like a solid object and more like a fluid. It ends either by silently fading from view or by exploding—sometimes violently enough to cause structural damage, injury, or death.
The Russian scientist A. I. Grigoriev analyzed more than 10,000 reported cases of ball lightning. Igor Stakhanov collected over 1,500 reports. Stanley Singer, François Arago, Camille Flammarion, and dozens of other researchers across two centuries have compiled, categorized, and analyzed eyewitness accounts. The phenomenon is not obscure. It’s not limited to a single culture or geography. The consistency across independent reports—the size, the duration, the luminosity, the movement patterns, the tendency to appear near lightning strikes, the capacity to pass through glass—is strong enough that the scientific community broadly accepts ball lightning as a real physical phenomenon. What it does not accept, because it doesn’t have one, is an explanation.
The competing hypotheses
There is no shortage of theories. There are dozens. None of them account for all the observed properties, and several of them are mutually exclusive.
The vaporized silicon hypothesis, advanced by John Abrahamson and James Dinniss at the University of Canterbury in 2000, proposes that when lightning strikes soil, it vaporizes the silica in the ground, separates the oxygen from the silicon dioxide, and produces a cloud of pure silicon nanoparticles. As the silicon recombines with atmospheric oxygen, it oxidizes and glows—a floating, burning aerosol bound together by its electrical charge. This hypothesis received significant support from the 2012 Lanzhou spectrum, which showed soil elements in the ball’s emission. Laboratory experiments have produced glowing balls lasting a few hundred milliseconds by evaporating pure silicon with electric arcs. The problem: those lab-produced balls are small, short-lived, and exist in partial atmospheres. They don’t drift through closed windows. They don’t last 10 seconds. They don’t appear inside aircraft.
The microwave bubble theory, proposed in 2017 by researchers at Zhejiang University in Hangzhou, suggests that at the tip of a lightning stroke reaching the ground, a relativistic electron bunch produces intense microwave radiation. The microwaves ionize the surrounding air and the radiation pressure evacuates the resulting plasma, forming a spherical plasma bubble that stably traps the microwave radiation inside it. The ball glows because the trapped microwaves continue to generate plasma. It fades when the radiation decays. It explodes when the structure destabilizes. This theory explains several otherwise puzzling properties: microwaves can pass through glass, which would account for ball lightning appearing indoors through windows. But verifying the theory experimentally would require hundreds of gigawatts of microwave power—about an order of magnitude beyond current laboratory capabilities.
The atmospheric maser-soliton theory, developed by Peter Handel, hypothesizes that the energy source is a large atmospheric maser—a region of air several cubic kilometers in volume where lightning creates a population inversion in the rotational energy levels of water molecules, generating coherent microwave radiation. Ball lightning appears as a plasma caviton at the antinodal plane of this radiation. The theory is elegant and explains the energy source problem—where does a small glowing sphere get enough energy to persist for 10 seconds?—but the atmospheric maser itself has never been directly detected.
The corona discharge theory, proposed by John Lowke at CSIRO in Australia, suggests that ball lightning is powered by the electrical field from dispersing charges in the earth after a lightning strike, producing a discharge similar to what occurs around high-voltage transformers. A 2024 paper proposed that ball lightning arises from a positive ion nucleus encased by a rotating shell of electrons whose motion stabilizes the plasma against collapse. Other researchers have proposed that ball lightning is a detached form of St. Elmo’s fire, or a self-trapped electromagnetic wave packet, or—in one preprint that a researcher described as having “that one attractive feature: that if the other end of the wormhole can go anywhere it wants, it might as well show up in somebody’s bedroom”—some kind of wormhole.
Why it resists explanation
The fundamental problem is data, not imagination. Ball lightning lasts seconds, appears unpredictably, and until the smartphone era left no physical trace that instruments could analyze. Almost everything we know comes from eyewitness reports—thousands of them, collected over centuries, from people who were not scientists, were not expecting to see anything unusual, and were often frightened. Eyewitness testimony is valuable for establishing that a phenomenon exists. It is nearly useless for determining physical mechanisms. Witnesses can estimate size and duration. They cannot estimate temperature, electromagnetic emission spectra, chemical composition, or internal structure.
The 2012 Lanzhou recording changed the empirical landscape, but marginally. One event, recorded at 900 meters distance, lasting 1.64 seconds. In July 2025, a couple in Rich Valley, Alberta, filmed what appeared to be ball lightning—a pale blue sphere hovering about seven meters above the ground for roughly 20 seconds after a lightning strike, moving with an oscillating quality. A 2025 paper in the Quarterly Journal of the Royal Meteorological Society analyzed this and other video evidence, noting that many purported ball lightning videos can be explained by power-line arcs, burning metallic debris, camera artifacts, or fireworks during storms. The authors described themselves as “skeptical believers”—convinced the phenomenon is real but unconvinced by most of the evidence offered to prove it.
The experimental situation is similarly constrained. Martin Uman at the University of Florida received US Air Force funding specifically to create ball lightning by triggering lightning strikes onto various materials. Of roughly 100 materials struck, four produced phenomena resembling ball lightning: a flame above salt water, glowing particles from silicon wafers, a persistent glow from a wet pine stump, and a glow hovering above a wet steel sheet. The team also tested bat guano, “for no reason except we had some lying around.” None of the results constituted a definitive reproduction.
What makes it a genuinely interesting problem
Ball lightning is one of the last atmospheric phenomena visible to the naked eye that lacks a consensus scientific explanation. We understand regular lightning. We understand tornadoes, waterspouts, St. Elmo’s fire, sprites, jets, and elves. We understand auroras. We understand rainbows. Ball lightning sits in a category with almost nothing else: observed frequently enough to be taken seriously, documented thoroughly enough to establish consistent properties, and resistant enough to explanation that dozens of competing theories coexist without any achieving dominance.
The resistance isn’t because the theories are bad. Several of them—particularly the vaporized silicon and microwave bubble models—are physically plausible and partially supported by evidence. The resistance is because the phenomenon itself seems to violate comfortable categories. A plasma that persists for seconds without an external energy source. A luminous object that passes through solid glass. A structure that can compress, reform, and explode. Each of these properties, individually, can be explained by at least one theory. No single theory explains all of them simultaneously.
The most productive framing might be that ball lightning isn’t one phenomenon. Different mechanisms—silicon oxidation, microwave trapping, corona discharge, maser effects—might each produce glowing spheres under different conditions, and the category “ball lightning” might be a folk taxonomy that groups visually similar but physically distinct events. If that’s the case, no unified theory will ever emerge because there’s nothing unified to theorize about. Multiple phenomena, one name, centuries of confusion.
We cover ball lightning alongside UAP sightings, cryptozoology, and other phenomena at the boundary between established science and the unexplained across our Fortean Phenomena course—including why the most honest answer to “what is ball lightning?” remains, after 200 years of scientific inquiry: we’re not sure, and we have 1.64 seconds of data.