This space engine breaks a law of physics. But a NASA test says it works anyway.

NASA scientists have been daydreaming about a new kind of engine that could carry astronauts to Mars in 70 days without burning any fuel. Now, in a new paper published in the peer-reviewed Journal of Propulsion and Power, they say that it might really work.

The paper, written by scientists at NASA’s Eagleworks Laboratories, tested a electromagnetic propulsion system, or “EM drive,” that generates a small amount of thrust simply by bouncing microwaves around a cone-shaped copper chamber. No propellant goes in, no exhaust comes out, and yet, somehow, the engine can make things move.

If you think that news sounds too good to be true, you’ve got good instincts — it just might be. This “impossible” fuel-less engine appears to violate one of the fundamental laws of physics.

Say what?

Hark back to your high school science classroom. Avert your eyes from theunfortunate hair styles and acne, if necessary, and try to focus on what’s written on the blackboard: For every action, there is an equal and opposite reaction.

That’s Newton’s third law of motion. It’s the principle that explains why pushing against a wall will send an ice skater zooming in the opposite direction. It also explains how jet engines work: As hot gases are expelled out the back of the plane, they produce a thrusting force that moves the plane forward.

But the EM drive doesn’t work that way. Its thrust seems to come from the impact of photons on the walls of the copper cavity. That would be like moving a car forward by just banging against the windshield.

And that works?

According to the new paper, yes. The Eagleworks scientists report that their machine generated 1.2 millinewtons of thrust per kilowatt of electricity pumped in. (That electricity could come from solar panels in a hypothetical spaceship.) That’s a fraction of thrust produced by the lightweight ion drives now used in many NASA spacecraft, National Geographic noted, but it’s a lot more than the few micronewtons per kilowatt produced by light sails, a proven technology that generates thrust using radiation from the sun.

Where did this idea come from?

The idea for an EM drive was first published a decade ago by British engineer Roger Shawyer. He argued that the drive isn’t really “reactionless” — instead, he argued, the thrust comes from radiation pressure. Microwaves inside the cavity create an imbalance of radiation that pushes against the walls and generates thrust.

The idea was hyped in headlines and splashed across the cover of New Scientist magazine, but most scientists were, and still are, extremely skeptical. There’s no theoretical explanation for how such an engine might work, and not all the possible sources of experimental error have been eliminated.

A team of scientists at China’s Northwestern Polytechnical University have beenworking to build their own EM drive, but their one positive result turned out to be a measurement error, according to the Christian Science Monitor. In 2014, independent inventor and chemical engineer Guido Fetta got the scientists at Eagleworks to evaluate his variation on the EM drive, which he calls Cannae. They concluded that it did produce a small amount of thrust, but didn’t speculate on what that might mean or what mysterious new laws of physics could have produced it. Fetta says he now wants to test the drive in space, according to Popular Mechanics.

NASA has been uncharacteristically restrained about this whole project. When asked about the Eagleworks experiments last year, the space agency told, “While conceptual research into novel propulsion methods by a team at NASA’s Johnson Space Center in Houston has created headlines, this is a small effort that has not yet shown any tangible results. NASA is not working on ‘warp drive’ technology.”

What does this new paper really mean?

The new finding does lend some credence to EM drive claims. It passed peer review, which means that several expert scientists reviewed the methodology and the results and found no major flaws. It also addressed one of the major knocks against past EM drive tests — that the engines heat up when activated, suggesting that hot air around the machines, rather than the photons inside them, might be what generates thrust. The Eagleworks scientists made sure this wasn’t the case by conducting their test in a vacuum.

This doesn’t mean that the Eagleworks EM drive definitely functions. Peer review is designed to make sure that studies are well designed and executed, and that the conclusions are reasonable — it’s not an endorsement. And plenty of findings published in solid scientific papers have later been found to be incomplete or incorrect. That’s how science is supposed to work: You draw conclusions based on the best evidence available, present them to your peers, and revise and refine as you conduct more tests and gather more data. The authors of the paper list nine possible sources of error in their experiment, and indicate that they need to do more tests to try to rule those out.

“The issue involved here is whether the experiment is seeing something real or not,” Jim Woodward, a physicist at California State at Fullerton, told Motherboard. “I know [co-author Paul March] does clean work and, to be honest, I suspect there may really be something there.”

But, Woodward added, there’s no theoretical explanation for the phenomenon that March and his colleagues report. That’s not necessarily disqualifying — evidence is evidence — but it is a good reason to stop and take a second look.