The study, Vineeth Chandran Suja confesses, was born in a moment of frustration.

As a graduate student at École Polytechnique, he had to design a research project for a class on biomechanics. He was drawing a blank—until he instinctively cracked his knuckles. “I thought, ‘This is something interesting and maybe fun,’” Suja says. Of all the studies published on knuckle cracking, no one had mathematically described exactly how the knuckle produces the crack. Suja thought it had to do with a drop in pressure inside the knuckle when it is stretched, so he decided to do the math.

While he chipped away at this idea, a paper came out in 2015 that seemed to overthrow 40 years of knuckle-cracking orthodoxy. “Mystery solved,” the headlines proclaimed. Scientists at the University of Alberta had put a man in an MRI machine while he cracked his knuckles. Stretching the joint caused the pressure inside to drop and a bubble to form. From the MRI, it looked like the sound came at the moment bubbles formed in the knuckles—not when the bubbles collapsed, as researchers previously thought.

But this did not make sense with Suja’s math. Suja—who is now a PhD candidate at Stanford—had not stuck anybody in an MRI machine, but he and two friends spent some time recording their knuckle cracks in an anechoic chamber, a special room that is built to be completely echo-free. The auditory signature of those recordings matched a mathematical model he had devised to describe how bubbles behave in the egg-white-like synovial fluid that lubricates joints. This model said that an audible crack should happen when the bubble is *partially* collapsing, but microbubbles may still remain.