2025MNRAS.538.2838K

The recent SH0ES determination of the Hubble constant, $H_0=73.04\pm 1.04$ km s$^{-1}$ Mpc$^{-1}$, deviates significantly by ${\approx }5\sigma$ from the Planck value, stimulating discussions on cosmological model extensions. To minimize statistical uncertainty and mitigate sensitivity to systematic errors in any single anchor distance determination, SH0ES combines Cepheids from various observations, including those from Type Ia supernova host galaxies, NGC 4258, and closer galaxies (MW, LMC, SMC, and M31), although this mixed sample may introduce unknown or subtle systematic errors due to comparing distant and closer Cepheids. To address this, we propose a subset excluding Cepheids from the closer galaxies, retaining only the NGC 4258 water megamasers as a single anchor, circumventing potential systematic errors associated with observational methods and reduction techniques. Focusing solely on these Cepheids yields $H_0=72.68\pm 1.67$ km s$^{-1}$ Mpc$^{-1}$, where the competitive statistical uncertainty of approximately $2.5{{\ \rm per\ cent}}$ is sufficient to identify a ${\approx }3\sigma$ discrepancy with the Planck $H_0$ value, while significantly mitigating Cepheid-related systematics associated with photometry determination, metallicity treatment, and dust-colour corrections. A period-bin analysis demonstrates that a global period-luminosity relation is unnecessary, challenging previous assumptions. We also validate the host Cepheid extinction estimates, showing their low reddening supports the use of optical photometry with systematic uncertainty smaller than the statistical uncertainty, potentially achieving higher precision than near-infrared photometry, given the lower optical background. We investigate potential explanations for the significant deviation from the Planck value observed in our analysis that do not involve Cepheids.