2013A&A...555A..88P

Optical quasar spectra can be used to trace variations of the fine-structure constant α. Controversial results that have been published in past years suggest that in addition to to wavelength calibration problems systematic errors might arise because of insufficient spectral resolution. The aim of this work is to estimate the impact of incorrect line decompositions in fitting procedures due to asymmetric line profiles.
Method. are developed to distinguish between different sources of line position shifts and thus to minimize error sources in future work.To simulate asymmetric line profiles, two different methods were used. At first the profile was created as an unresolved blend of narrow lines and then, the profile was created using a macroscopic velocity field of the absorbing medium. The simulated spectra were analysed with standard methods to search for apparent shifts of line positions that would mimic a variation of fundamental constants. Differences between position shifts due to an incorrect line decomposition and a real variation of constants were probed using methods that have been newly developed or adapted for this kind of analysis. The results were then applied to real data. Apparent relative velocity shifts of several hundred meters per second are found in the analysis of simulated spectra with asymmetric line profiles. It was found that each system has to be analysed in detail to distinguish between different sources of line position shifts. A set of 16 FeII systems in seven quasar spectra was analysed. With the methods developed, the mean α variation that appeared in these systems was reduced from the original Δα/α=(2.1±2.0_stat)x10^–5 to Δα/α=(0.1±0.8_stat)x10^–5. We thus conclude that incorrect line decompositions can be partly responsible for the conflicting results published so far.