EDGES versus SARAS 3: Two Experiments, One Sky, Opposite Answers, Priced

EDGES reported a sky-averaged 21-cm absorption trough at redshift 17.21 centred near 78 MHz with depth about 0.50 K; SARAS 3 reported a null rejecting the EDGES best-fit profile at 95.3 per cent confidence. Deriving the gas temperature from adiabatic cooling after decoupling at redshift 150, standard cosmology permits a maximum depth of 265 mK, so EDGES exceeds the standard maximum by a factor of 1.89. An excess radio background requires 88.4 K at that epoch, 1.78 times the CMB there (2.13 K equivalent today, far below the extrapolated ARCADE-2 level), at the price of roughly 1000 times the radio efficiency of present-day galaxies. Baryon-dark-matter cooling requires cooling the gas from 5.99 K to 3.36 K via a velocity-to-the-minus-fourth cross-section near 1e-21 cm2, a tuned corner that evades CMB anisotropy bounds only through that velocity dependence. An instrumental artefact requires only a 500 mK foreground residual, 0.01 per cent of the roughly 5000 K foreground, with structure on a 9.5 MHz scale. Verdict: all three readings remain open but at very different prices; the artefact reading is cheapest and is supported by the SARAS 3 null; the discriminating observation is a third fully independent global measurement cross-checked against an interferometric 21-cm power spectrum at the same redshift. All derived numbers are produced by the deposited compute.py. AI use is disclosed: this paper was produced with the assistance of Dave, an AI system, under the direction of the named author. William Alexander Aldred is sole author of record.