1998A&A...333..665D

Previous studies of visible-region high-excitation lines of GaII in HgMn stars have usually concluded that these lines (λλ4251-4262 and λ6334) yield abundance estimates for gallium ∼1dex greater than the UV resonance lines. Of the explanations proposed in the literature, we find that the presence of hyperfine structure (hfs) in the lines is the most likely. We analyse Lick Hamilton Echelle CCD spectra by spectrum synthesis with the code UCLSYN. Using the Bidelman & Corliss (1962ApJ...135..968B) measurements of hfs in GaII, and the calculations given by Lanz et al. (1993A&A...272..465L), with the oscillator strengths by Ryabchikova & Smirnov (1994AZh....71...83R), we determine a difference in abundance (visual-UV) of only 0.2dex. Of this difference, about 0.1dex can be explained by the simplified approximation to the true hfs, which can be estimated theoretically, and the remaining 0.1dex can probably be accounted for by the stratification of Ga found by Smith (1995A&A...297..237S, 1996Ap&SS.237...77S). We conclude that the visible-region and UV abundances are in agreement within the errors of the determinations, and that the anomaly previously found by several investigations is an artifact of the simplified atomic line structures assumed, or results from the use of very different gf-values from those adopted here.