Warm Molecular Hydrogen in Metal-poor Compact Galaxies: A Uniform JWST/MIRI Archival Analysis

Abstract Pure-rotational H 2 lines trace warm molecular gas, but they do not determine the total molecular reservoir. We uniformly reanalyze public JWST/MIRI/MRS spectroscopy of I Zw 18, Leo P, SBS 0335-052 E, and CGCG 007-025. The analysis distinguishes reproducible product-level warm-H 2 measurements from multiphase inferences that require a common spatial footprint. We calibrate line-flux uncertainties and detection S/N from local residuals while accounting for positive spectral autocorrelation and test that calibration with residual-preserving injections. A machine-readable decision table records 32 target–transition outcomes: five accepted X1D detections, 24 clean empirical upper limits, and three quality-control cases retained for review and quality assurance. No target retains three transitions; we therefore report upper-level populations without fitting an excitation slope or inferring a warm-H 2 mass. For Leo P, VLA data give a beam-scale H i detection, but the broad-beam test has insufficient MIRI coverage and the better-covered smaller-beam test does not yield a statistically significant H i signal. Neither branch meets the joint requirements for a source-matched H i mass, so no numerical warm-H 2 /H i ratio is reported. The quality-controlled inventory, local-noise calibration, and footprint audit provide a reproducible foundation for future source-matched multiphase observations.