The UVES Large Program for testing fundamental physics I. Bounds on a change in α towards quasar HE 2217−2818

Context. Absorption-line systems detected in quasar spectra can be used to compare the value of the fine-structure constant, α, measured today on Earth with its value in distant galaxies. In recent years, some evidence has emerged of small temporal and also spatial variations in α on cosmological scales. These variations may reach a fractional level of ≈ 10 ppm (parts per million). Aims. To test these claims we are conducting a Large Program of observations with the Very Large Telescope’s Ultraviolet and Visual Echelle Spectrograph (UVES), and are obtaining high-resolution (R ≈ 60 000) and high signal-to-noise ratio (S/N ≈ 100) UVES spectra calibrated specifically for this purpose. Here we analyse the first complete quasar spectrum from this programme, that of HE 2217−2818. Methods. We applied the many multiplet method to measure α in five absorption systems towards this quasar: zabs = 0.7866, 0.9424, 1.5558, 1.6279 , and 1.6919. Results. The most precise result is obtained for the absorber at zabs = 1.6919 where 3 Fe ii transitions and Al ii λ1670 have high S/N and provide a wide range of sensitivities to α. The absorption profile is complex with several very narrow features, and it requires 32 velocity components to be fitted to the data. We also conducted a range of tests to estimate the systematic error budget. Our final result for the relative variation in α in this system is Δα/α = +1.3 ± 2.4stat ± 1.0sys ppm. This is one of the tightest current bounds on α-variation from an individual absorber. A second, separate approach to the data reduction, calibration, and analysis of this system yielded a slightly different result of −3.8 ppm, possibly suggesting a larger systematic error component than our tests indicated. This approach used an additional 3 Fe ii transitions, parts of which were masked due to contamination by telluric features. Restricting this analysis to the Fe ii transitions alone and using a modified absorption profile model gave a result that is consistent with the first approach, Δα/α = +1.1 ± 2.6stat ppm. The four other absorbers have simpler absorption profiles, with fewer and broader features, and offer transitions with a narrower range of sensitivities to α. They therefore provide looser bounds on Δα/α at the ≳10 ppm precision level. Conclusions. The absorbers towards quasar HE 2217−2818 reveal no evidence of any variation in α at the 3-ppm precision level (1σ confidence). If the recently reported 10-ppm dipolar variation in α acr