One of the most remarkable examples of spectrum lines which are common in celestial bodies, but which have hitherto resisted all attempts to reproduce them in the laboratory, is afforded by the higher members of the Balmer series of hydrogen. As many as 29 members of this series have been observed by Dyson and Evershed in the chromosphere of the sun, but the greatest number observed in the laboratory by Ames and by Cornu was only 13, and the last of these were of such a character that it would hardly have been possible to record them without a previous knowledge of their localisation. In many respects the failure to reproduce in the laboratory lines whose chemical origin is known, and which are so prominent in celestial spectra, is even more conspicuous than in the case of the nebular and coronal lines, which cannot yet be referred to any atom known in chemistry, and which may be due to substances which do not, or perhaps cannot, exist under terrestrial conditions. A further interest has been added to the problem by the important theoretical wrork of Bohr, whose theory of the production of the Balmer series requires that the space occupied by a hydrogen atom, in the process of emitting lines of the higher members of the Balmer series, is such that these radiations cannot be expected to be visible except under conditions of extremely low pressure. Bohr has pointed out that this view is consistent with the appearance of the lines in celestial spectra, and our inability to produce them under the conditions ordinarily obtaining in the laboratory. Liveing and Dewar have found that in a mixture of the more volatile gases of the atmosphere, consisting mainly of neon and helium and containing hydrogen, the Balmer series could be traced as far as the ninth member; and in a recent investigation we have made a quantitative comparison of the distribution of intensity in the earlier members of the series in hydrogen, and in neon containing hydrogen as an impurity. This observation of Liveing and Dewar is remarkable, but we have recently found that similar results can be obtained in helium containing hydrogen at pressures so great that, on the theoretical considerations above referred to, it would seem impossible that these radiations should be detected.