Constraints on additional relativistic particles. Top: evolution of the effective degrees of freedom for Standard Model particle density, g_*, as a function of photon temperature in the early Universe. Vertical bands show the approximate temperature of neutrino decoupling and the QCD phase transition, and dashed vertical lines denote some mass scales at which corresponding particles annihilate with their antiparticles, reducing g_*. The solid line shows the fit of Borsanyi et al. (2016) plus standard evolution at T_γ <  1  MeV, and the pale blue bands the estimated ±1σ error region from Saikawa & Shirai (2018). Numbers on the right indicate specific values of g_* expected from simple degrees of freedom counting. Bottom: expected ΔN_eff today for species decoupling from thermal equilibrium as a function of the decoupling temperature, where lines show the prediction from the Borsanyi et al. (2016) fit assuming a single scalar boson (g = 1, blue), bosons with g = 2 (e.g., a massless gauge vector boson, orange), a Weyl fermion with g = 2 (green), or fermions with g = 4 (red). One-tailed 68% and 95% regions excluded by Planck TT,TE,EE+lowE+lensing+BAO are shown in gold; this rules out at 95% significance light thermal relics decoupling after the QCD phase transition (where the theoretical uncertainty on g_* is negligible), including specific values indicated on the right axis of ΔN_eff = 0.57 and 1 for particles decoupling between muon and positron annihilation. At temperatures well above the top quark mass and electroweak phase transition, g_* remains somewhat below the naive 106.75 value expected for all the particles in the Standard Model, giving interesting targets for ΔN_eff that may be detectable in future CMB experiments (see e.g. Baumann et al. 2018).