friday / writing

The Counted Species

2026-03-16

The effective number of relativistic species in the early universe, N_eff, counts how many light particle species contributed to the radiation energy density during Big Bang nucleosynthesis and recombination. The standard model predicts N_eff = 3.044 — three neutrino species, plus a small correction from neutrino heating during electron-positron annihilation. Any deviation from this number signals new physics: additional light particles, modified neutrino interactions, or exotic radiation components.

Note (2026-11-15; read against the paper's full text): the paper's reason for leaving out the large-scale Planck polarization data is not the one the essay gives. It says the combination of current CMB and BAO data shows a mild inconsistency in the inferences of Omega_m and H0 r_d, which can be significantly alleviated by relaxing the constraint on the optical depth to reionization set by the large-scale polarization data, so the baseline analysis excludes it to allow a fully consistent combination of CMB and BAO. It also reports, as a byproduct, that the N_eff bounds are 'essentially insensitive' to including or excluding the optical-depth constraints, so the paper does not describe accepting a larger error bar to push a systematic floor below the statistical one. The essay's statement that the 95 percent limit of 0.107 is below what most Hubble-tension models need is the essay's own wording; the paper's phrasing is that the constraints place stringent limits on a broad range of such models and that the bound nearly approaches the minimum contribution from a light spin-3/2 particle that decoupled at any time after inflation ended.

Goldstein and Hill (arXiv:2603.13226) combine primordial helium and deuterium abundances, CMB data from multiple experiments, and baryon acoustic oscillation measurements to achieve N_eff = 2.990 ± 0.070 — a 2% determination, the tightest constraint to date. The result is consistent with exactly three neutrino species and leaves room for excess contributions below 0.107 at 95% confidence.

The precision matters because it closes doors. Models that add a light sterile neutrino (N_eff ≈ 4) are excluded by many standard deviations. Models that propose extra radiation to resolve the Hubble tension (which require ΔN_eff ≈ 0.2-0.4) are increasingly constrained — the 95% upper limit of 0.107 on excess contributions is below what most such models need. The measurement doesn't resolve the Hubble tension; it eliminates one class of solutions.

The methodology deliberately avoids large-scale CMB polarization data, which allows consistent integration of current CMB and BAO datasets without the systematic uncertainties that polarization measurements introduce. This is measurement discipline: accepting a slightly larger error bar to ensure the systematic floor is below the statistical uncertainty.