Line shifts due to blending
The reported investigation is concerned with the case in which the measured position of a strong line of the spectrum is blended with a weaker one, taking into account the traditional solution proposed by Hartmann (1901) concerning the mean of the solar lines. An arbitrary element regarding the use of the centroid is related to the decision at which separation the components pass from being 'blended' to being 'resolved'. At this separation, the adopted wavelength suddenly jumps from the centroid of the blend to the position of the stronger line alone. It is pointed out that this discontinuous behavior is physically implausible. Attention is given to a solution involving a gradual diminution of the effect of the weaker line with increasing separation. Thus the weight of each component should depend on its position as well as its intensity. When partially blended lines cannot be decomposed into their components, and an effective line-position for the hole blend is needed, the line peak is the criterion least affected by blending. A method is presented for calculating such peak positions.