The UVES Large Program for testing fundamental physics - III. Constraints on the fine-structure constant from three telescopes

Large statistical samples of quasar spectra have previously indicated possible cosmological variations in the fine-structure constant, a. A smaller sample of higher signal-to-noise ratio spectra, with dedicated calibration, would allow a detailed test of this evidence. Towards that end, we observed equatorial quasar HS 1549+1919 with three telescopes: the Very Large Telescope, Keck and, for the first time in such analyses, Subaru. By directly comparing these spectra to each other, and by 'supercalibrating' them using asteroid and iodine-cell tests, we detected and removed long-range distortions of the quasar spectra's wavelength scales which would have caused significant systematic errors in our α measurements. For each telescope we measure the relative deviation in α from the current laboratory value, Δα/α, in three absorption systems at redshifts zabs = 1.143, 1.342 and 1.802. The nine measurements of Δα/α are all consistent with zero at the 2σ level, with 1σ statistical (systematic) uncertainties 5.6-24 (1.8-7.0) parts per million (ppm). They are also consistent with each other at the 1s level, allowing us to form a combined value for each telescope and, finally, a single value for this line of sight: Δα/α = -5.4 ± 3.3stat ± 1.5sys ppm, consistent with both zero and previous, large samples. We also average all Large Programme results measuring Δα/α = -0.6 ± 1.9stat ± 0.9sys ppm. Our results demonstrate the robustness and reliability at the 3 ppm level afforded by supercalibration techniques and direct comparison of spectra from different telescopes.