## Abstract

A new attempt has been made to resolve spectroscopically the fine structure of the $\alpha $ line, 6563 angstrom, of the Balmer series of heavy hydrogen. A deuterium discharge tube, whose walls were partly of copper, was immersed in liquid hydrogen and run at very low current densities. By this means the Doppler width of the individual lines was sufficiently reduced to enable the satellite between the main components of the 'doublet' to be resolved much more completely and to be measured more accurately than was possible before. The spectroscopic resolution was achieved with a Fabry-Perot interferometer of large diameter, mounted internally in a prism spectrograph. The measurements support the assumption that the 2S$_{\frac{1}{2}}$ term is higher than is predicted by the Dirac theory, and lead to the value 0$\cdot $0369 $\pm $ 0$\cdot $0016 cm.$^{-1}$ for the amount of the shift, in good agreement with the latest value found by Lamb & Retherford from microwave measurements. The relative widths of the main components and the interval between them also provide qualitative evidence for the shift of the 2S$_{\frac{1}{2}}$ term. The intensities of the lines were found to agree approximately with the prediction of the Dirac theory. The weak line 2P$_{\frac{3}{2}}$-3S$_{\frac{1}{2}}$ was partly resolved from the neighbouring strong line 2P$_{\frac{3}{2}}$-3D$_{\frac{5}{2}}$. The provisional value for the distance between them would indicate that the 3S$_{\frac{1}{2}}$ term is shifted very little, if at all. Further attempts are being made to obtain a more reliable value for this term.