The primordial helium abundance has begun drifting below its theoretical anchor. Standard BBN with the CMB baryon density predicts a helium mass fraction Y_p = 0.247 with sub-percent precision, and for decades measurements in metal-poor dwarf galaxies, extrapolated to zero metallicity, obligingly agreed. The EMPRESS survey changed the picture: deep Subaru spectroscopy of extremely metal-poor galaxies, the cleanest extrapolation anchors available, yielded Y_p = 0.2370 +0.0034/-0.0033 (EMPRESS VIII, 2022, arXiv:2203.09617), roughly 3 sigma below the standard prediction, and the 2025 follow-up with near-infrared helium lines and 29 galaxies confirms a moderately low Y_p = 0.2402 +/- 0.0040 (EMPRESS XV, arXiv:2506.24050). A plateau that was a pillar of concordance has shifted downward at the precision frontier.
If the low value is physical, the standard model's cleanest fix is a primordial lepton asymmetry: an electron-neutrino degeneracy parameter xi_e near 0.05 shifts neutron-proton freeze-out enough to lower Y_p, but this imports an asymmetry orders of magnitude larger than the baryon asymmetry with no mechanism, touching N_eff and the Hubble tension in correlated ways (the EMPRESS analysis itself derives xi_e = 0.05 +0.03/-0.02). The conservative alternative blames systematics in the helium emissivity modeling, the metallicity extrapolation, or underlying stellar absorption, each under active investigation, with the helium lines' sensitivity to density and temperature structure the perennial suspect.
The standing is a live 2-to-3-sigma anomaly in the element least forgiving of error: helium's prediction is nearly parameter-free, so a confirmed low plateau forces new neutrino-sector physics into the first second, and upcoming 30-meter-class spectroscopy of the EMPRESS anchors will decide whether the shift survives.