Two studies published September 28, 2026, report that CHIME detected the cosmological 21-centimeter signal at 12.4σ in selected observations from 2019. The measurement is an autocorrelation power spectrum of neutral-hydrogen emission fluctuations across 608.2–707.8 MHz, corresponding to redshifts 1.34–1.01. It measures the signal’s spatial pattern, not the universe’s total hydrogen inventory. (Detection paper; interpretation paper)
What CHIME measured
A power spectrum describes how strongly a signal varies across different spatial scales. CHIME’s autocorrelation analysis calculated that pattern from the radio data themselves, without requiring an external galaxy survey. The reported result covers wavenumbers of about 0.4–1.5 h Mpc⁻¹, scales the companion study describes as nonlinear.
The 12.4σ figure is the detection significance for that full-band power spectrum under the study’s analysis. The team also reported detections in each of the two sub-bands:
| Measurement band | Frequency | Redshift | Reported significance |
| Full band | 608.2–707.8 MHz | 1.34–1.01 | 12.4σ |
| Lower-frequency sub-band | 608.2–658.2 MHz | Centered near 1.24 | 9.1σ |
| Higher-frequency sub-band | 658.2–707.8 MHz | Centered near 1.08 | 8.6σ |
How the analysis isolated the signal
The sample comprised 94 selected sidereal days from 2019. The researchers rejected eight of 102 candidate days after data-quality checks and used nighttime observations, when the Sun was below the horizon.
The detection paper describes improvements to radio-frequency-interference detection and masking, achromatic beamforming, and foreground filtering before time averaging. These steps helped the team extract the radio map’s own spatial power spectrum for the autocorrelation measurement. (CHIME detection study)
What the simulation comparison adds
The companion paper interprets the measured hydrogen clustering through an effective amplitude, 𝒜HI². It reports a value of 3.55, with statistical uncertainties of +0.96 and −1.32 and a systematic uncertainty of ±0.61. The parameter combines neutral-hydrogen abundance, hydrogen bias and redshift-space effects, so it is not an abundance measurement by itself.
The paper also gives a conditional abundance estimate: ΩHI × 10³ = 0.79, with statistical uncertainties of +0.20 and −0.16. That estimate uses a literature-based Gaussian prior for hydrogen bias, with a mean of 1.56 and a standard deviation equal to 20% of the mean. (Interpretation paper)
CHIME’s measured power differs from the IllustrisTNG TNG100 and TNG300 predictions by 3.1σ and 4.0σ, respectively. The authors point to nonlinear hydrogen clustering and redshift-space distortions as likely areas of mismatch, without identifying a single cause. The comparison concerns simulated neutral-hydrogen power at the measured scales.
What the result says about cosmology
The two studies focus on detecting and interpreting neutral-hydrogen clustering. The companion analysis adopts a background cosmology based on Planck-derived ΛCDM parameters; the studies report no measurement of how dark energy evolves.
The interpretation paper says CHIME had nearly seven years of archived data and that analysis of the full archive would be presented in future work. It gives no publication date for that analysis.
