On the effect of asymmetry on wave-length determinations

It is a familiar experience in spectroscopy that the actual accuracy which is attained in precise measurements of wave-length is almost invariably disappointing, and falls far short of the precision to be expected from a consideration of the probable errors. In particular, the measurement of unsymmetrical lines, difficult in itself, is affected by displacements which depend on the limited resolving power of the spectrograph. It is no doubt well known that this effect must be operative, and it has been considered, for example, by Grebe and Bachem in their search for the Einstein effect in the solar spectrum. But we are not aware of any precise formulation of its magnitude, although a close approximation to the apparent displacement of the maximum of an unsymmetrical line, expressed in the terms of the resolving power and the asymmetry, can be obtained from simple considerations. We have at present no very precise knowledge of the intensity distribution curves of the unsymmetrical lines which are found in the spectra of the electric arc and of numerous other sources. But we have convinced ourselves, from an inspection of the photometric curves obtained by King and Koch and St. John and Babcock, that for the present purpose, we may assume, without serious error, that the lines may be represented by the purely artificial device of combining two error curves.