Same apparatus design, different G: the NIST versus BIPM discrepancy
In plain words
NIST rebuilt the instrument used by the international measurement bureau (BIPM) and got a value about 2 parts in 10000 lower. Since the design is the same, the difference should be traceable to a specific cause.
Precise statement
The NIST measurement (2026, Metrologia), using a torsion-strip balance of the BIPM design in both servo and Cavendish modes, gives $G = 6.67387e-8\,\mathrm{cm}^3\,\mathrm{g}^{-1}\,\mathrm{s}^{-2}$, about 235 ppm below the BIPM 2013 value. Identify the effect (mass metrology, torsion-strip anelasticity, source-mass density inhomogeneity, alignment, environment) that accounts for the difference.
What would settle it
A cross-check with exchanged source masses or fibers, or a third realization of the same apparatus, that isolates the component responsible for the 235 ppm offset.
Status in the literature
Unverified note
NIST's blinded result was unblinded in July 2024 and published in April 2026.