How the Sun Tugs at Us: Testing the Universality of Free Fall express

Original: "An Improved Torsion Balance Test of the Equivalence Principle Towards the Sun"
arXiv:2602.02815 · 2026-02-02 · CC BY 4.0 · 1 min
Scientists tested whether the Sun pulls equally on different materials and found only agreement with expectations — to four decimal places.

On a lab table sits an instrument that looks like a balance, but instead of pans it has two weights made of different metals — beryllium and aluminum. The balance is suspended from a thin fiber that twists if the weights are pulled differently. The whole apparatus turns to follow the Sun, like a sunflower tracking the light, but instead of light it catches the Sun's gravity.

The Sun pulls on every object on Earth: a stone, a feather, a piece of metal. The question is whether it pulls different materials with equal strength.

If the Sun pulled beryllium slightly harder than aluminum, the fiber would twist. But the instrument saw no difference. This means the pull does not depend on what a body is made of — exactly as general relativity predicts. Spacetime curvature acts on everything equally, just as a pencil and a hammer fall together in an elevator when the cable snaps.

The physicists pushed the precision to the limit: if any difference in the pull existed, it would be less than two parts in a trillion. That is like spotting a difference in weight between two kilogram masses if one were lighter by the weight of a single grain of sand. Earlier experiments looked at other sources of gravity, but now the Sun itself, the most massive object nearby, has been tested. The new result is four times more precise than previous solar measurements and 20% more precise than any other torsion-balance experiment.

The torsion balance was first used by Henry Cavendish in 1798 to weigh the Earth. Since then, it has become the go-to instrument for probing the subtlest effects of gravity.

\eta_{\odot, Be-Al} \leq 2.1 \times 10^{-13}
The equivalence-principle violation parameter for the Sun, measured with beryllium and aluminum test bodies, does not exceed 2.1×10⁻¹³ (95% confidence limit).
Scientists
Albert EinsteinRobert H. DickeIsaac NewtonJohn MichellRobert HookeJacob Bekenstein
Concepts
torsion balance equivalence principle test bell inequality Lorentz Invariance Violation measurement error parity violation equipartition theorem proportionality error threshold measurement accuracy cp symmetry