The writer is engaged in a program of observing eclipsing binaries with the principal aim of obtaining masses and radii of stars of various types. For this purpose it is necessary that radial velocities of both components of a system be obtained. Since in many systems the fainter component is also cooler, it is advantageous to work at longer wavelengths than has been usual. These notes consist principally of preliminary results of the observational program. The grating spectrographs of the 100-inch (coude) and 60-inch (Cassegrain) telescopes on Mount Wilson can be used effectively in the visual region of the spectrum as well as in the photographic region. The dispersions employed range from 10 to 20 Â/mm in the photographic region and from 15 to 40 Â/mm in the visual region. Particular emphasis has been placed on observations of the D lines since it is the D lines of a fainter, cooler component that are most likely to be observable. Because of the great strength and breadth of Ha in the spectra of the hotter components and because of its tendency to appear in emission in close binary systems, Ha is a less favorable line for the purpose. The magnitude limit (three-hour exposure in average seeing) at the D lines is about 8.5 with dispersions of 15 and 20 A/mm with the 100-inch and about 8.0 with dispersions of 30 and 40 A/mm with the 60-inch. In work of this nature, particularly in the investigation of the lines of a faint component, the highest available dispersion should be used. It has been my experience that where series of spectrograms with different dispersions are available, those of lower dispersion add but little weight to results obtained with higher dispersion. Stars north of declination +50° cannot, unfortunately, be observed with the 100-inch coude spectrograph.