By Iris Kowalczyk · Published May 8, 2026 · Updated May 8, 2026
Last reviewed: May 8, 2026.
Direct Answer
Ball lightning is a rare luminous sphere reported during or just after thunderstorms. Witnesses describe a glowing ball, roughly grapefruit-sized, lasting seconds before fading or exploding. Atmospheric physicists treat the phenomenon as real but unexplained. The leading hypotheses involve oxidizing silicon nanoparticles, microwave cavity discharges, and plasma solitons. None has closed the case.
What the Witness Sheets Say
A cold-case file begins with the witness sheet. Ball lightning has thousands of those. Folklorist W. Morris drew the first modern compilation in the 1930s. Physicist Mark Stenhoff catalogued more than 200 first-hand accounts after a 1976 Nature note drew correspondence from across the British Isles [1]. The Royal Meteorological Society and TORRO have continued to log reports since. Survey numbers vary, but a working figure is that several percent of adults claim to have seen one [2].
The descriptions converge with unusual discipline for a folkloric phenomenon. Diameters cluster between ten and twenty centimeters [3]. Most balls last seconds, sometimes a half minute, rarely longer. Color reports skew red, orange, yellow, white. Movement is often horizontal, at walking pace, sometimes drifting against the wind. Sound is intermittent: hissing, buzzing, or silence. The ball ends in one of two ways. It fades quietly. Or it terminates in a sharp report and a smell of sulfur and ozone [4].
Demographic patterns sharpen the pattern further. Reports cluster in the immediate aftermath of cloud-to-ground strikes. They cluster in rural or wooded settings more than in dense cities, though urban accounts exist. They cluster in summer in temperate latitudes and in monsoon seasons in subtropical ones. The witness population skews toward farmers, climbers, sailors, and aircrew — populations that spend hours under open sky during electrified weather. None of these patterns proves a mechanism. Each one constrains the candidate list.
The 1638 Widecombe Report
The earliest well-documented English case is the Great Thunderstorm at the parish church of St Pancras in Widecombe-in-the-Moor, Devon, on 21 October 1638. Two contemporary tracts describe a fiery ball striking the church during the afternoon service. Four parishioners died, dozens were injured. The masonry was scorched and a pew was torn loose. Modern atmospheric physicists, including Stenhoff, treat the account as a candidate ball-lightning event rather than a settled identification. The records are clergy and printed broadsheet; the physics interpretation is provisional.
The witness pool widens after that. A 1753 case in Saint Petersburg killed Russian physicist Georg Wilhelm Richmann while he attempted to measure atmospheric electricity using a vertical iron rod. Contemporaries described a pale blue ball leaping from the rod to his forehead. The death warned a generation of natural philosophers off unguarded experiments. Richmann’s notes survive at the Russian Academy of Sciences. Whether the ball was the same phenomenon witnesses report today, or a related discharge from his apparatus, remains an open question in the file.
Building the Timeline of Theories
A working investigator builds the timeline of explanations the way one builds a suspect timeline. Names, dates, claims, what would distinguish them.
The eighteenth century treated the phenomenon as a curiosity. Benjamin Franklin and the Royal Society correspondents noted it without theorizing closely. The serious scientific attempt arrived later. In 1955, Soviet physicist Pyotr Kapitsa proposed that ball lightning is a glow discharge sustained by microwave standing waves channeled along ionized air from the parent cloud [5]. Kapitsa’s paper appeared in Doklady Akademii Nauk while he was rebuilding his standing after political dismissal. The model required a microwave generator above; critics noted that thunderstorms do not reliably produce such fields.
The hypothesis-list grew through the late twentieth century. Plasma vortex models. Antimatter speculation. Charged dust aggregates. Standing-wave masers. By the time of the 1997 International Symposium on Ball Lightning, Stenhoff could enumerate more than a dozen distinct mechanisms then under serious discussion [1]. None had decisive evidence. The phenomenon was rare, brief, and hostile to instrumentation.
Two interim positions deserve note. The skeptic’s reading, voiced occasionally in atmospheric-physics journals, was that ball lightning reports might be retinal afterimages — bright flashes from a nearby strike persisting in the visual cortex as a drifting bright spot. This explanation accounts for some short-duration cases. It does not account for multi-witness events, photographs, scorched fixtures, or the 2014 spectrum. The opposite position, voiced from the engineering side, was that the phenomenon was so well-attested in field reports that the missing piece was instrumentation, not existence. The 2014 capture vindicated the engineers.
The Silicon-Vapor Hypothesis
In February 2000, John Abrahamson and James Dinniss of the University of Canterbury, New Zealand, published a paper in Nature that became the most widely cited modern model. They argued that when ordinary lightning strikes silicate-rich soil, it vaporizes silica and reduces it with carbon to elemental silicon nanoparticles. The particles are ejected in a chained, fluffy network. As they oxidize back to silica in air, they release stored chemical energy slowly, producing a luminous, buoyant ball [6]. The model accounts for color, size range, lifetime, and the soil-strike correlation. Bench-top burning of pure silicon spheres reproduced some of the optical signatures, though not the floating, drifting ball as a whole [7].
The chemistry is straightforward in outline. A direct strike injects energies on the order of a gigajoule into a small volume of soil. Temperatures spike. Silica reduces to silicon in the presence of carbonaceous matter — roots, humus, charcoal. The molten silicon rapidly fragments into nanoparticles as the plasma cools. Surface area governs the rest. A loose chain of nanoparticles oxidizes more slowly than a solid pellet, releasing the stored chemical energy over the seconds-to-tens-of-seconds window the witness sheets describe. The model also predicts dependence on soil type. Sandy quartz-rich soils should produce more silicon-rich balls than clay or peat. The 2014 Qinghai event, on alpine plateau soil rich in silicates, fits the prediction.
The 2014 Spectroscopic Capture
For decades the case suffered from the chain-of-custody problem any cold-case investigator recognizes. Photographs existed. Spectra did not. Most camera records were grainy, unscaled, and uncorroborated by a second witness. Several were exposed as misidentified St Elmo’s fire, glow discharges, or fraudulent composites. The first natural-event spectroscopic record arrived by accident.
On 23 July 2012, Jianyong Cen, Ping Yuan, and Simin Xue of Northwest Normal University in Lanzhou were operating slitless spectrographs and high-speed video on the Qinghai Plateau, observing ordinary cloud-to-ground lightning. A ball of light, roughly five meters in apparent diameter, appeared after a strike, drifted about fifteen meters, and dissipated within 1.6 seconds. The spectrograph recorded its emission lines [8].
The spectrum showed silicon, iron, and calcium — the elemental signature of the local soil. The team published the result in Physical Review Letters in January 2014. APS Physics framed the paper as the first natural ball-lightning spectrum on record [9]. The correspondence with the Abrahamson–Dinniss prediction was direct: a soil strike produced a luminous globe whose composition matched the strike substrate. The result did not close the case. It tightened it.
What the Laboratory Has and Has Not Done
Three laboratory programs have produced ball-lightning-like plasmoids under controlled conditions, none of them a fully fielded match.
At the Max Planck Institute for Plasma Physics in Garching, researchers generate buoyant plasmoids over an aqueous electrolyte by a moderate current pulse. The objects last around one hundred milliseconds, are spherical, and detach from the electrode [10]. The Russian physicist Egely produced similar globules earlier in the development of this technique. The objects are short-lived for ball lightning standards but reproducible.
A second route uses microwaves to eject plasmoids from silicon substrates. Mitchell, Dikhtyar, and Jerby reported buoyant fireballs ejected from a silicon target inside a microwave cavity, with lifetimes and emission consistent with the silicon-oxidation model [11]. A third line, reported in Nature Photonics in 2026, generates relativistic terahertz solitons at a tungsten tip in a supersonic argon jet, sustaining ball-lightning-like coherence over roughly one hundred nanoseconds — orders of magnitude shorter than reported events but a clean demonstration of self-sustaining electromagnetic spheres in plasma [12].
A working investigator weighs evidence by what each line of work would and would not predict. The water-electrolyte plasmoids reproduce the visual appearance and detachment but not the typical lifetime, and they require an explicit electrode geometry not present in open atmosphere. The microwave-silicon route reproduces the chemical signature and a buoyant fireball but requires a microwave cavity. The terahertz soliton work reproduces stability mathematics but is too short-lived and too narrowly conditioned to map onto the field reports. Each is a partial match. Each leaves part of the witness sheet unexplained.
The reconstructions overlap but do not converge. Burning-silicon models explain composition and lifetime. Microwave-cavity and soliton models explain stability and field structure. Witness reports demand both: composition consistent with soil chemistry, and stability not easily explained by chemistry alone. A 2024 paper in arXiv even proposes a magnetic-monopole signal as a fringe alternative; mainstream atmospheric physics has not adopted it, and the proposal is filed in the notes column rather than the report.
Why the Case Stays Open
A cold-case file closes when a new piece of evidence forces a single reconstruction. Ball lightning is not there yet.
The phenomenon resembles a series of unsolved disappearances more than it resembles a single homicide. Each report is its own small case. Each has a separate witness, a separate location, a separate atmospheric context. The pattern emerges from the aggregate, not from any single record. Stenhoff’s collection of more than two hundred first-hand accounts functions like a missing-persons archive: similar circumstances, similar physical descriptions, no closed file [1]. The 2014 spectrum is the equivalent of a recovered piece of physical evidence in one of those files. It does not solve the others. It tells the working investigator what to look for in the next one.
The phenomenon is rare and unrepeatable to schedule. Field instruments capable of catching one are seldom pointed at the right square kilometer in the right second. The Qinghai capture was statistically lucky and methodologically narrow — one ball, one spectrum [9]. Replication is the next step. Several groups, including the Royal Meteorological Society, are running citizen-witness collection programs to gather better-conditioned reports, including video [13].
The hypotheses themselves are not equally constrained. Silicon nanoparticle oxidation has direct laboratory analogs and one matched-spectrum field event. Microwave cavity models have produced laboratory plasmoids but lack a verified field correlate. Soliton models are mathematically elegant and now have terahertz laboratory demonstrations, but the path from a hundred nanoseconds to twenty seconds in open air is unsettled.
The honest position from the witness sheets and the spectra is this. Ball lightning happens. The 2014 capture established that a real, optically luminous, locally composed sphere followed a lightning strike. The mechanism is narrowing. Two or three reconstructions still fit. What would distinguish them is a second spectrum from a different soil type, a confirmed microwave field at the parent cloud, or a laboratory plasmoid that survives more than a few seconds without containment. The investigation is no longer about whether the phenomenon exists. It is about which model the next clean record forces us to keep.
A short procedural note for readers who carry phones into thunderstorms. If a luminous sphere appears nearby, a thirty-second video at standard frame rate, with the camera held steady against a fixed reference, captures more usable data than any single eyewitness account. Note the GPS coordinates, the time, the apparent diameter relative to a known object in the frame, the color, and the manner of termination. Submit the file to the Royal Meteorological Society or to a university atmospheric physics group. The next confirming spectrum, or the next disconfirming one, may be a phone video. Cold cases close on details that warm ones swallow.
Frequently Asked Questions
Is ball lightning real?
Atmospheric physicists treat ball lightning as a real but rare phenomenon. The 2014 Cen-Yuan-Xue spectroscopic capture in Physical Review Letters provided the first natural-event spectrum, recording silicon, iron, and calcium consistent with the local soil [8][9]. Witness counts run into the thousands across two centuries.
How long does ball lightning last?
Most reported events last between one second and roughly thirty seconds, with a typical duration near twenty to twenty-five seconds [3]. Some accounts describe minute-long events, but those are rare and harder to verify.
How big is ball lightning?
Reported diameters span one to one hundred centimeters. The most common size is ten to twenty centimeters — roughly a grapefruit to a bowling ball [3]. The 2014 Qinghai event was apparently larger, around five meters in apparent diameter at observation distance.
What color is ball lightning?
Witnesses most often report red, orange, yellow, or white. Roughly sixty percent of cases describe red-orange-yellow tones; about a quarter describe white. Blue, green, or color-shifting reports are less common [3].
What causes ball lightning?
No mechanism is settled. The leading hypotheses are silicon nanoparticle oxidation following soil strikes (Abrahamson and Dinniss, 2000), microwave cavity discharges (Kapitsa, 1955), and plasma soliton models. The 2014 spectrum favors the silicon-oxidation route but does not exclude others [5][6][8].
Has ball lightning been recreated in the laboratory?
Several laboratories have produced plasmoid analogs. The Max Planck Institute for Plasma Physics generates buoyant plasmoids over water electrolytes; microwave-silicon experiments produce ejected fireballs; recent terahertz-soliton experiments demonstrate stable coherent spheres at nanosecond scale [10][11][12]. None reproduces the full duration of reported field events.
What was the 2014 ball lightning study?
Researchers from Northwest Normal University, observing ordinary lightning on the Qinghai Plateau in July 2012, accidentally captured a ball-lightning event on slitless spectrographs and high-speed video. Their analysis appeared in Physical Review Letters in January 2014 and showed soil-matched emission lines [8][9].
Is the 1638 Widecombe-in-the-Moor event ball lightning?
The Great Thunderstorm at St Pancras Church on 21 October 1638 is one of the earliest documented candidate events. Two contemporary printed tracts describe a fiery ball entering the church during service. Modern researchers treat the case as plausible but not confirmed; the records are clergy and broadsheet, not instrumental [1].
Can ball lightning pass through walls?
Witness reports of ball lightning passing through closed windows or aircraft fuselages exist but are contested. Some can be explained by the ball forming on one side and a corresponding charge effect on the other, others by misidentification. No verified instrumented record of wall-penetration exists [1][2].
How dangerous is ball lightning?
Most close-encounter accounts describe heat, sound, and a sulfurous smell rather than direct injury. The 1638 Widecombe event killed four parishioners, but the exact mechanism — ball lightning or conventional lightning strike inside the church — is not settled. Death by ball lightning specifically is rare in the historical record.
Sources and Further Reading
- [1] Stenhoff, M. (1999). Ball Lightning: An Unsolved Problem in Atmospheric Physics. Springer/Plenum.
- [2] Royal Meteorological Society. Ball Lightning. rmets.org/metmatters/ball-lightning
- [3] Wikipedia summary of physical characteristics, drawing on NASA and peer-reviewed surveys.
- [4] Singer, S. (1971). The Nature of Ball Lightning. Plenum Press.
- [5] Kapitsa, P. L. (1955). On the Nature of Ball Lightning. Doklady Akademii Nauk SSSR, 101, 245-248.
- [6] Abrahamson, J. and Dinniss, J. (2000). Ball lightning caused by oxidation of nanoparticle networks from normal lightning strikes on soil. Nature, 403, 519-521. nature.com/articles/35000525
- [7] Paiva, G. S. et al. (2007). Production of ball-lightning-like luminous balls by electrical discharges in silicon. Physical Review Letters, 98, 048501.
- [8] Cen, J., Yuan, P. and Xue, S. (2014). Observation of the Optical and Spectral Characteristics of Ball Lightning. Physical Review Letters, 112, 035001. link.aps.org/doi/10.1103/PhysRevLett.112.035001
- [9] APS Physics. (2014). First Spectrum of Ball Lightning. physics.aps.org/articles/v7/5
- [10] Max Planck Institute for Plasma Physics. Ball lightning research. ipp.mpg.de/2977926/kugelblitze
- [11] Mitchell, J. B. A., Dikhtyar, V. and Jerby, E. (2013). Observations of Ball-Lightning-Like Plasmoids Ejected from Silicon by Localized Microwaves. Materials, 6(9), 4011-4030.
- [12] Boerner et al. (2026). Ball-lightning-like relativistic terahertz solitons. Nature Photonics. nature.com/articles/s41566-026-01899-y
- [13] Eos / AGU. Have You Seen Ball Lightning? eos.org
Explore more open files in the parent collection: Unsolved Mysteries and Enigmas.
