An envelope can be scientific equipment if the thing inside it is a number nobody is allowed to know.
For roughly a decade, physicist Stephan Schlamminger and colleagues at the National Institute of Standards and Technology worked to measure the universal gravitational constant, big G. One of Schlamminger's colleagues had quietly altered a crucial input by a secret amount. The correction stayed sealed away while the team collected data, found mistakes, and decided which adjustments were justified.[2]
This was not sabotage. It was a defense against getting the answer they wanted.
The pull of a familiar answer
Big G sets the strength of gravity in Newton's equation. Gravity may run planets, galaxies, dropped mugs, and most other objects with no regard for our paperwork, but measuring its strength in a laboratory is unusually difficult. Recent high-precision measurements disagree by hundreds of parts per million, far more scatter than physicists see for many other fundamental constants.[3]
The NIST team was trying to replicate a measurement made at the International Bureau of Weights and Measures in France. It used the same torsion balance, an instrument that detects the tiny twist produced when movable masses attract one another. The earlier experiment had reported one of the highest recent values of G.[3]
That made agreement dangerous in a very ordinary way. If you know the destination, every borderline correction can begin to look like the road toward it. Nobody has to falsify anything. A reasonable person can simply spend longer investigating errors that move the result away from expectation and feel satisfied sooner when a correction moves it back.
So Patrick Abbott, a NIST colleague, subtracted a number known only to him from measurements of the experiment's masses. Schlamminger could analyze the scrambled result, but he could not see whether it matched the French value until the hidden correction was restored.[2]
The envelope even survived a false ending. Schlamminger was ready to open it in 2022, then stopped after finding an overlooked air-pressure effect. He returned to the analysis and finally unsealed the number during a conference talk on July 11, 2024.[2] This is an excellent use of suspense, although perhaps not one that conference organizers should apply to the lunch menu.
The envelope worked; the numbers did not
The revealed result was G = 6.67387 × 10−11 cubic meters per kilogram per second squared, with a relative uncertainty of 57 parts per million.[1] It agreed with the world average within one standard deviation, but it was about 250 parts per million below the result from the French experiment it was meant to replicate.[3]
The team tried the apparatus with copper masses and then sapphire ones, and those results were virtually identical.[2] It also examined possible trouble from temperature imbalances, mass-density variations, and other effects. No single culprit was large enough to explain the difference. Schlamminger's judgment, reported by Physics, was that several smaller causes were probably involved.[3]
That is less glamorous than announcing a crack in Newtonian gravity, and the paper treats unknown instrumental effects as the more plausible explanation.[1] But the disappointment is the useful part. Repeating the apparatus did not repeat the answer. The experiment exposed a limit in how reliably this family of measurements can be reproduced.
The blinding did exactly what it was supposed to do. It did not improve the torsion balance, remove heat from the room, or identify every background force. It kept one especially persuasive force out of the analysis: the wish to agree.
Science is often described as a machine for producing answers. Here it behaved more like a machine for preserving an honest question. Ten years of work ended with a careful measurement, a stubborn mismatch, and no satisfying villain. That counts as progress, though it is admittedly terrible merchandise for a gift shop.
The envelope did not contain the value of gravity. It contained permission for the result to be disappointing.