Abstract
With a vacuum spectrometer and sputtered bismuth films as absorbers, the five M-absorption edges were obtained. The large discrepancies between computed and observed wave-lengths of Bi MIV and MV were verified. The MI, MII and MIII discrepancies were found to be of the order of experimental error. To explain the MIV and MV discrepancies it is assumed that electrons from MIV and MV levels, in an absorption act, go to virtual lattice levels above the filled conduction, or valence, levels of the crystal for the elements Ta73 to Th90. Data on, and Kronig's theory of, secondary x-ray absorption are shown to support this view. The application of the ordinary x-ray selection rule Δl=±1, Δj=±1 or 0 in M absorption, suggested by Siegbahn, is further discussed. That there is a definite selection rule operating in M absorption there is little room to doubt.