On the night of November 12, 1988, residents around Quebec's Saguenay region watched glowing orbs drift silently across the sky. Some hovered. Some pulsed. Some raced along ridgelines like will-o'-the-wisps. Eleven days later, a magnitude 6.0 earthquake tore through the same area. For centuries, reports like these — luminous plumes, flickering sheets, spherical lights bobbing before, during, or after tremors — were dismissed as folklore, mass hysteria, or misidentified aircraft. Then geologists started taking them seriously.

The phenomenon is called earthquake light, and it has been documented since at least 373 BCE, when Greek writers described "immense columns of fire" preceding the destruction of Helike. Pliny the Elder catalogued them. Chinese seismological records mention them. A 1965 photograph from Matsushiro, Japan finally gave the scientific community physical evidence that these weren't just tall tales.

The problem is nobody knows exactly what causes them. Several competing hypotheses:

• Piezoelectricity — quartz-rich rocks under massive tectonic stress generate voltage, the same effect that makes your grill lighter click into flame
• Triboelectric effects — friction between rock faces separates charges, the same mechanism that shocks you on a doorknob
• Positive hole activation — a newer theory from NASA's Friedemann Freund suggests that stressed peroxy defects in igneous rocks release charge carriers that flood upward through the crust, ionizing the air when they hit the surface
• Fracture-electrification — the actual breaking of rock bonds releases electrons

If you've ever peeled sticky tape off a surface in a dark room and watched it glow faintly blue, you've witnessed triboluminescence — a cousin phenomenon. Wintergreen Life Savers famously spark when you bite them for the same reason. Now imagine that same charge separation happening across cubic kilometers of crust grinding against itself, and the scale becomes plausible.

What makes earthquake lights particularly strange is their variety. Witnesses report sheet lightning without clouds, blue flames licking the ground, glowing balls that persist for tens of seconds (real lightning lasts milliseconds), and even coronal discharges around metal fences and power lines hours before a quake. During the 2009 L'Aquila earthquake in Italy, someone caught 4-second flashes on video that couldn't be explained by transformer failures, because the power was already out.

Freund's team analyzed 65 well-documented cases and found something unnerving: earthquake lights cluster around rift environments and subvertical faults, not the transform faults where most quakes happen. That means they might actually be a precursor signal, appearing minutes to weeks before the shaking starts — which would make them the closest thing seismology has ever had to a warning system.

Skeptics remain, and the physics is still contested. But the next time someone tells you the sky glowed before an earthquake, they might not be crazy. They might just be watching gigatons of stone slowly turning into a battery.