Abstract Previous, large samples of quasar absorption spectra have indicated some evidence for relative variations in the fine-structure constant (Δα/α) across the sky. However, they were likely affected by long-range distortions of the wavelength calibration, so it is important to establish a statistical sample of more reliable results from multiple telescopes. Here we triple the sample of Δα/α measurements from the Subaru Telescope which have been ‘supercalibrated’ to correct for long-range distortions. A blinded analysis of the metallic ions in six intervening absorption systems in two Subaru quasar spectra provides no evidence for α variation, with a weighted mean of Δα/α = 3.0 ± 2.8stat ± 2.0sys parts per million (1σ statistical and systematic uncertainties). The main remaining systematic effects are uncertainties in the long-range distortion corrections, absorption profile models, and errors from redispersing multiple quasar exposures on to a common wavelength grid. The results also assume that terrestrial isotopic abundances prevail in the absorbers; assuming only the dominant terrestrial isotope is present significantly lowers Δα/α, though it is still consistent with zero. Given the location of the two quasars on the sky, our results do not support the evidence for spatial α variation, especially when combined with the 21 other recent measurements which were corrected for, or resistant to, long-range distortions. Our spectra and absorption profile fits are publicly available.