The air suddenly feels different. Dogs begin to howl, and the horses become restless. A burst of bright light, and seconds later, the ground starts to rumble and shake.

Earthquake precursors — from strange animal behavior to temperature anomalies — may sound like urban legends, but new findings have lent some scientific credence to one phenomenon called earthquake lights. The aerial flashes of light seen before or during temblors may have to do with large jumps in electrical activity as the earth cracks open.

The results were presented at the American Physical Society’s March Meeting in Denver on Thursday by Rutgers University biomedical engineer Troy Shinbrot. His lab has created a miniature model of earthquake-like jamming and cracking, and has found huge voltage jumps that result from the shifting of granular material used to mimic the earth.

He used tanks filled with different types of grains, from kitchen flour to glass beads, and moved them relative to one another in a quick start-and-stop motion to create cracks. A non-contact voltage probe focused on measuring a certain crack. The voltages differed from material to material, but the overall pattern remained the same. When the grains split open, they measured a positive voltage spike, and when the split closed, a negative spike.

Sometimes the voltage ­changes were as high as hundreds of volts — but as of now, the researchers cannot explain why.

Electrical spikes of 100 volts or more can result when a crack opens and closes in a bed of powder under stress in a laboratory-scale experiment. (Troy Shinbrot/Rutgers University/APS)

“Our first suspicion was that this has to be a mistake,” Shinbrot said at a news conference Tuesday.“We did many tests to try to rule out these spurious effects, and so far we have failed.”

Although at times mistaken for UFO sightings, observations of earthquake lights date back hundreds of years. But most reports are for quakes with high magnitudes, of 5 or greater on the Richter scale.

Boston College seismologist John Ebel, who studies earthquake forecasting and patterns, has reviewed reports of electromagnetic precursors to quakes. Generally, the timing, strength and duration of the observed lights vary widely — but he is nevertheless “comforted” by these recent findings of Shinbrot and his colleagues.

“I personally think that there are legitimate observations of lightning-like dis­charges — electrical glows of some sort — that occur before earthquakes,” said Ebel, who was not involved in the study. “There’s too much evidence to discount their existence.”

Shinbrot referenced other light phenomena linked to making and breaking surface contact: Mercury glows as it slides on glass, Scotch Tape lights up as it is ripped off the roll and Wint O Green Life Savers mints create a spark when bitten. “These are effects that occur when cracks open,” Shinbrot said. “All of the data is really fascinating and unexplained.”

The examples he mentioned show off a phenomenon called triboluminescence, or instances where energy is released in the form of light through the forced breaking of chemical bonds, such as when a solid is crushed, rubbed or scratched.

“We know that electricity, among other things, helps hold materials together,” Ebel said. “When you break bonds, you change the electrical properties of the system, which could cause a cascade.”

Shinbrot’s lab originally was looking at the effects of electrostatics on pharmaceutical powders and how a charge can make them stick to surfaces. He had read about earthquake lights, and because he had the powders and the right equipment, he decided to try a related experiment. The team first tried using the pharmaceutical powders they already had on hand, such as acetaminophen, bu later moved on to unbleached wheat flour and special glass beads that can visualize areas of stress.

They also found that the voltage ­changes seem to be associated with ­changes in ­stresses from surfaces in contact rather than solids actually breaking apart as in some triboluminescent examples.

But many questions remain. While earthquake lights are sometimes reported days before the event, the experiment clearly shows electrical jumps with each crack and uncrack. Also, the actual measured currents with each movement are tiny.

Physicist Friedemann Freund of the NASA Ames Research Center in California nevertheless applauded the work: “The observation is really well done, the experiments are well conducted, but the full understanding is not yet available.”

Freund is conducting experiments this summer that he hopes will pull together all known information about earthquake lights in a way that makes sense given Shinbrot’s results — but he isn’t prepared to talk about them just yet.

Ebel thinks their model could translate well to areas with fault gouge — rock with a very small grain size, formed by tectonic movement along a fault zone — such as parts of California. But in Quebec, some earthquakes originate from depths of almost 20 miles where rock could even be liquefied.

“In a lab experiment, you can constrain conditions, whereas in the earth, things aren’t constrained,” he said. “You don’t know what the grain size is or the distribution of grains.”

He also wonders whether these results will hold at scales ranging not in inches, but spanning tens of miles. Shinbrot is in the midst of testing effects of scale using larger tanks of grain. Also, he wants to measure voltages at more places than just at the split to see whether there is a certain multi­dimensional pattern of electricity.

Kim is a freelance science journalist based in Philadelphia.