to the star the spectroscope could give us no information as to the motion. The motion to or from the earth would be invariable. To show the result of the orbit being seen obliquely, let E be the earth and A S be the plane of the orbit seen edgewise. Drop the perpendicular A M upon the line of sight. Then, while the star is moving from S to A the spectroscope will measure the motion as if it took place from S to M. Since S M is less than A S, the measured velocity will always be less than the actual velocity, except in the rare case when the plane of the orbit is directed toward the earth. Since the spectroscope can give us no information as to the inclination under which we see the orbit, it follows that the actual orbital velocities of the spectroscopic binaries must remain unknown. We can only say that they cannot be less, but may be greater to any extent than that shown by our measures.
Fig. 4. The Mills Spectrograph of the Lick Observatory.
If the components of a binary system do not differ greatly in brightness, its character may be detected without actually measuring the radial velocities. Since the motion is shown by a displacement of the spectral lines and since, in any binary system, the two components must always move in opposite directions, it follows that the displacements of the spectral lines of the two stars will be in opposite directions. Hence, when one of the stars, say A, is moving toward us, and the other, say D, from us, all the spectral lines will appear double, the lines made by A being displaced toward the blue end of the spectrum and those by B toward the red end. After half a revolution the motion will be reversed and the lines will again be double; only the lines of star A will now be on the red side of the others. Between these two phases will
