Two prisms were formed by splitting a thick plate of glass diagonally.  These gave such good results that two others were made in the same way, and the entire battery of four prisms is ordinarily used.  The safety and convenience of handling the prisms is greatly increased by placing them in square brass boxes, each of which slides into place like a drawer.  Any combination of the prisms may thus be employed.  As is usual in such an investigation, a great variety of difficulties have been encountered, and the most important of them have now been overcome.

4. Faint Stellar Spectra.—­The 28 inch reflector will be used for the study of the spectra of the faint stars, and also for the fainter portions near the ends of the spectra of the brighter stars.  The form of spectroscope mentioned above, in which the collimator and slit are replaced by a concave lens, will be tried.  The objects to be examined are, first, the stars known to be variable, with the expectation that some evidence may be afforded of the cause of the variation.  The stars whose spectrum is known to be banded, to contain bright lines, or to be peculiar in other respects, will also be examined systematically.  Experiments will also be tried with orthochromatic plates and the use of a colored absorbing medium, in order to photograph the red portions of the spectra of the bright stars.  Quartz will also be tried to extend the images toward the ultra-violet.

5. Absorption Spectra.—­The ordinary form of comparison spectrum cannot be employed on account of the absence of a slit.  The most promising method of determining the wave lengths of the stellar spectra is to interpose some absorbent medium.  Experiments are in progress with hyponitric fumes and other substances.  A tank containing one of these materials is interposed and the spectra photographed through it.  The stellar spectra will then be traversed by lines resulting from the absorption of the media thus interposed, and, after their wave lengths are once determined, they serve as a precise standard to which the stellar lines may be referred.  The absorption lines of the terrestrial atmosphere would form the best standard for this purpose if those which are sufficiently fine can be photographed.

6. Wave Lengths.—­The determination of the wave lengths of the lines in the stellar spectra will form an important part of the work which has not yet been begun.  The approximate wave lengths can readily be found from a comparison with the solar spectrum, a sufficient number of solar lines being present in most stellar spectra.  If, then, satisfactory results are obtained in the preceding investigation, the motion of the stars can probably be determined with a high degree of precision.  The identification of the lines with those of terrestrial substances will of course form a part of the work, but the details will be considered subsequently.