Aims. We present a detailed analysis of a very high resolution (R ≈ 112 000) spectrum of the quasar HE 0515−4414 obtained using the High Accuracy Radial velocity Planet Searcher (HARPS) mounted on the ESO 3.6 m telescope at the La Silla observatory. The main aim is to use a HARPS spectrum of very high wavelength calibration accuracy (better than 1 mÅ), to constrain the variation of α ≡ e 2 / c and investigate any possible systematic inaccuracies in the wavelength calibration of the UV Echelle Spectrograph (UVES) mounted on the ESO Very Large Telescope (VLT). Methods. A cross-correlation analysis between the Th-Ar lamp spectra obtained with HARPS and UVES is carried out to detect any possible shift between the two spectra. Absolute wavelength calibration accuracies, and how that translates into the uncertainties in ∆α/α are computed using Gaussian fits for both lamp spectra. The value of ∆α/α at z abs = 1.1508 is obtained using the many multiplet method and simultaneous Voigt profile fits of HARPS and UVES spectra. Results. We find the shift between the HARPS and UVES spectra has a mean around zero with a dispersion of σ 1 mÅ. This is shown to be well within the wavelength calibration accuracy of UVES (i.e. σ 4 mÅ). We show that the uncertainties in the wavelength calibration induce an error of about ∆α/α ≤ 10 −6 in determining the variation of the fine-structure constant. Thus, the results of non-evolving ∆α/α reported in the literature based on UVES/VLT data should not be heavily influenced by problems related to wavelength calibration uncertainties. Our higher resolution spectrum of the z abs = 1.1508 Damped Lyman-α system toward HE 0515−4414 reveals more components compared to the UVES spectrum. Using only Fe ii lines of the z abs = 1.1508 system, we obtain ∆α/α = (0.05 ± 0.24) × 10 −5 . This result is consistent with the earlier measurement for this system using the UVES spectrum alone.