The hypothesis, proposed by Druyvesteyn and others, that the $K\alpha$-satellites arise from transitions of a $2p\left(L_{\mathrm{I}\mathrm{I},\mathrm{I}\mathrm{I}\mathrm{I}}\right)$ electron into the $1s \mathrm{}\left(K\right)\mathrm{}$ shell in the presence of one or two further vacancies in the $L$ shell is tested (1) by comparing in the case of sodium the separations from the parent line for the lines due to transitions of the first type calculated by self-consistent field methods with the observed separations of the satellites and (2) by effecting a similar comparison between observation and theory for a number of other elements, the separations being obtained in this case by a simplified method previously used by H. C. Wolfe but extended to include the effects of spin-orbit interaction. It is found that the results can be accounted for if the lines ${\alpha }^{\prime }$, ${\alpha }_3$ and ${\alpha }_4$ are attributed to transitions in atoms lacking one additional electron in the $L$ shell, probably of the type $1s2p\to {\mathrm{}\left(2p\right)\mathrm{}}^2$, and the lines ${\alpha }_5$, ${\alpha }_6$, ${\alpha }_7$ and ${\alpha }_8$ to transitions in atoms lacking two additional electrons in the $L$ shell. A consideration of the effect of chemical combination supports these conclusions.

• Received 5 October 1934

DOI:https://doi.org/10.1103/PhysRev.46.1040