How different are the Liège and Hamburg atlases of the solar spectrum?
Context: The high-fidelity solar spectral atlas prepared by Delbouille and co-workers (Liège atlas), and the one by Neckel and co-workers (Hamburg atlas), are widely recognised as the most important reference spectra of the Sun at disc-centre in the visible wavelength range. Both datasets serve as fundamental resources for many researchers, in particular for chemical abundance analysis. But despite their similar published specifications (spectral resolution, noise level), the shapes of spectral lines in the two atlases differ significantly and systematically. Aims: Knowledge of any instrumental degradations is imperative to fully exploit the information content of spectroscopic data. We seek to investigate the magnitude---and to explain the possible sources---of these differences. We provide the wavelength-dependent correction parameters that need to be taken into account when the spectra are to be compared with e.g. synthetic data. Methods: A parametrically degraded version of the Hamburg spectrum is fitted to the Liège spectrum. The parameters of the model (wavelength shift, broadening, intensity scaling, intensity offset) represent the different characteristics of the respective instruments, observational strategies and data processing. Results: The wavelength scales of the Liège and Hamburg atlases differ on average by 0.5 mÅ with a standard deviation of $\pm$2 mÅ, except for a peculiar region around 5500 Å. The continuum levels are offset by up to 18% below 5000 Å but stably at a 0.8% difference towards the red. We find no evidence for spectral straylight in the Liège spectrum. Its resolving power is almost independent of wavelength but limited to about 216000, between two to six times less than specified. When accounting for the degradations determined in this work, the spectra of the two atlases agree within a few parts in 10$^3$.