Another systematic effect in the determination of stellar abundances

A small systematic error that can affect the determination of stellar abundances is described. When the structure of the atmosphere of a star is represented by a line-blanketed model, the general line opacity should also be included in the calculation of the equivalent width of each individual line. The failure to do this makes the line too deep, and a given, observed equivalent width is matched by an abundance that is less than the correct value. Tests of the dependence of this systematic effect on line strength, wavelength, surface gravity, effective temperature, and spectral resolution are presented. The error can be avoided by analyzing each line by spectral synthesis, by using a portion of an opacity distribution function to represent general line blanketing in the computation of each line profile, or by making a careful differential abundance analysis.