The Number Physics Cannot Pin Down: NIST Measures Big G

Drop a ball. It falls. That is gravity, the most familiar force in your life. Yet the single number that tells us exactly how strong gravity is, called the gravitational constant, is the worst-known number in the rulebook of physics. A team at the National Institute of Standards and Technology, working with a borrowed instrument from the world's metrology bureau in Paris, just spent eight years trying to nail it down. Their new value, published in Metrologia, made the puzzle bigger, not smaller.

A Constant That Will Not Sit Still

Newton wrote down his gravity formula in 1687. Plug in two masses and the distance between them, multiply by G, and you get the pull. Simple. Every other constant in physics, like the speed of light or the charge of an electron, is now known to nine or ten decimal places. The gravitational constant is barely known to four.

Worse, when different labs measure it, they get different answers. Their error bars do not even overlap. Either gravity is doing something strange, or every lab has a hidden mistake nobody has caught yet.

Why Gravity Is the Hardest Force to Measure

Hold two cricket balls a foot apart. The gravity between them is about a billionth of a billionth of their weight. A car driving past outside, a person walking into the room, even warm air rising from a lamp can swamp the tiny signal you are trying to read.

That is the problem. Gravity from any nearby mass is dwarfed by noise from everywhere else. So labs use heavy weights, ultra-quiet rooms, and tools so sensitive they can pick up a twist smaller than the width of an atom.

How Do You Weigh Gravity With a Twisted Ribbon?

The NIST team used a tool first dreamed up at the International Bureau of Weights and Measures in Paris. It is a flat metal disk hung from a thin copper ribbon, like a heavy plate dangling from a strip of tape.

Four small copper weights sit on the plate. Four big copper weights sit on a turntable around it. Move the big weights, and their gravity tugs the plate sideways. The ribbon twists by a sliver of a degree. Measure that twist, and you can work out G.

To check themselves, the team built two ways of reading the twist. One lets the plate swing freely. The other uses tiny electric pushes to hold the plate still and reads off how much push was needed. Both methods should give the same G. If they do not, something is wrong.

The Hidden Heat That Fooled Everyone

Halfway through the project, the team noticed something odd. The value of G they were measuring crept higher whenever the vacuum chamber had a little more leftover gas inside it. Gravity should not care about air pressure. Something else was going on.

They traced it to heat. A tiny temperature gap of a few millionths of a degree across the chamber was enough to nudge the leftover gas molecules in one direction. That gentle nudge pushed on the plate and faked a gravity signal. They call this a thermophoretic force, gas pushed by warmth.

This was a quiet killer. No previous team had caught it. The NIST group fixed it by moving the vacuum pump further away, wrapping the whole rig in foam, and measuring at two pressures so they could subtract the effect.

What This Mystery Says About Science Itself

The team did not improve the world's knowledge of G. They made the cloud of disagreement a little wider. But they also did something most measurements never do. They wrote up every dead end, every fooled gauge, every dimple in a copper cylinder that turned out to matter.

One of their copper weights was slightly three-sided instead of perfectly round. That tiny shape error, smaller than a hair's width, was enough to bias the answer. So was a quirk in the angle-reading telescope, which they sent to the Physikalisch-Technische Bundesanstalt in Germany for a special calibration.

This is what honest precision physics looks like. Not a clean victory, but a stack of hard-won lessons that the next team will need.

• The number is real, the value is not settled: Half a century of measurements still disagree on G beyond their stated uncertainties.
• Heat fakes gravity: A temperature difference of millionths of a degree can produce a false signal as big as the real one.
• Hardware does not equal answer: The same machine in two labs gave two different values, which means the room and the team matter as much as the instrument.

"Precision metrology is not merely about converging on a number. It is about the rigorous exposure of unknowns." — Schlamminger et al., Metrologia, 2026.