THE SPECTRA OF DOUBLE STARS.—A note on “The Discovery of Double Stars by means of their Spectra” is contributed by Prof. E. C. Pickering to Astronomische Nachrichen, No. 3034. When the components of a close binary system have similar spectra, relative orbital motion in the line of sight may cause periodic doubling of the lines. But if the spectra be not similar any lines common to both ought to be conspicuously strong, and, provided the components have not equal and opposite velocities in the line of sight, ought also to be displaced with reference to other lines. Thus, if one component of a close binary system has a Group V. spectrum, like our sun, and the other a Group IV. spectrum, in which strongly marked hydrogen lines is the main feature, the resulting spectrum will have a composite character, and careful measurements should show that the position of the hydrogen line is periodically displaced when compared with the lines characteristic of the solar-type spectrum α Canis Majoris is the brightest star having this composite spectrum, and the wave-length of the hydrogen line G, derived from a comparison with three lines of greater and three lines of smaller wave-length, was 434˙09, which exceeds that derived from the solar spectrum by 0˙03. Similar measures of the hydrogen line h gave a save-length of 410˙22, which also exceeds that in the solar spectrum by 0˙03. From this displacement it would appear that if the phenomenon is due to the relative motion of a faint component, it is receding at the rate of 20 kilometres persecond, as compared with the bright component. An examination shows that the following stars have the composite spectrum referred to: γ Andromedæ, H.P. 650, є Boötis, α Scorpii, and β Cygni, all of which are known to be double; also π Persei, ζ Aurigæ, δ Sagittarii, 31 Cygni, and β Capricorni. In the cases of the last two, the spectra of the distant companions are distinctly separated from those of the chief stars. Although the strong hydrogen lines in the spectra investigated may be due to the presence of a faint companion, their intensity may also be due to many other causes. Thus, the strong hydrogen lines in the solar spectrum are not due to the integration of the spectrum of the sun and that of a companion. It is necessary, therefore, to determine whether the displacement is subject to a periodic variation or not, in order to test this method of discovering close binaries.