What was measured is the beam's speed

Zhang, Soter and colleagues, in a peer-reviewed Nature Physics paper dated 14 September 2026, report a muonium beam extracted into vacuum from a thin layer of superfluid helium at the Paul Scherrer Institute (PSI) in Switzerland. Muonium is a Coulomb-bound positive anti-muon and electron; its mass is dominated by the muon, a second-generation lepton. In the setup a μ+ beam at p μ = 12.5 MeV c−1 comes to rest at a mean depth of 35 μm in helium-II at T ≈ 200 mK, captures an electron and diffuses toward the surface.[1]

The fit gives a longitudinal speed v = 2180 m s−1 (+160/−130) and a diffusion speed v diff = 26 m s−1 (+6/−4). Velocity spreads σ v ≈ 80–150 m s−1 are consistent with the data at 90 percent confidence; the ~150 m s−1 upper bound is enough to read the beam as superthermal (Δ v ≪ v). The vacuum conversion efficiency is η Mu = 8.2 percent (+0.8/−1.1). The authors write that the yield is comparable to the densest diffuse sources while the longitudinal velocity distribution is an order of magnitude narrower than the coldest thermal sources at 100 K. Those figures describe the beam's phase space.[1]

The search space moved; the acceleration has not

The paper recalls that the weak equivalence principle has so far been tested only with neutral composites of first-generation particles — atoms, neutrons, antihydrogen. Muonium could open a free-fall test with a second-generation lepton in the absence of the strong interaction. The authors write that the beam is expected to enable muonium interferometry and a per-cent-level measurement of gravitational acceleration. The same text projects Δ a / g ≈ 0.01 relative precision in ~100 days at an expected R Mu ≈ 1 × 10^5 s−1 at PSI. The paper's closing arithmetic for a three-grating interferometer uses the measured v and η Mu.[1]

A quieter reading is that the superthermal speed enlarges interference contrast, while grating alignment, partial coherence and the muon lifetime can still lock g Mu out. The observation that would move the argument is a count of gravity-shifted interference fringes in a vertical geometry. The same brightness also opens 1S–2S spectroscopy for bound-state quantum electrodynamics and the muon mass; that line remains a promised second use of the beam. The next score is whether a fringe shift in vertical geometry can be counted against the Δ a / g ≈ 0.01 target.[1]

The muon bill sits on another table

At the end of July this column wrote that the calculation which closed a 25-year muon-moment gap had sent the bill to the pion measurements. The beam reported here leaves that tension in place. That path runs through hadronic vacuum polarization; this source carries the muon into vacuum, off a superfluid surface, at a measured speed. The two jobs look at the same particle: one asks what feeds the calculation, the other asks for the instrument of a fall and of a 1S–2S line.[1], [2]