the sun's image be formed by mirrors instead of lenses. Thus the telescope should be a horizontal reflector of the cœlostat type. The aperture of the mirrors should be as great as possible, since the high dispersion of the spectroheliograph, even in the most favorable cases, will involve long exposures.
But the most important requisite is such a condition of the atmosphere as will give the finest possible definition of the solar image. This would involve the establishment of the instruments at some particularly favorable site, where careful telescopic tests have shown the definition of the solar image to be exceptionally tine. Such a site is most likely he found in regions where the sky remains cloudless for weeks at a time: a point of great importance, since at such a place the various phases of changing phenomena could be followed up day after day with the same instruments. A still more thorough study of constantly varying solar phenomena could be secured through the cooperation of a chain of suitably equipped observatories, so distributed in longitude as to permit the sun to be kept continuously under observation.
With such large spectroheliographs, employed with a well-defined solar image of sufficient size, it should be possible to study not only the vapors lying around and above sun-spots, but those which constitute the spot itself. With sufficiently high dispersion, for example, it ought to be possible to secure photographs with the slit set at different points on some of the broadest of the 'widened' lines, giving sections of the vapors of the dark umbra of the spot at different depths below the surface. A comparative study of the forms of the umbra, as recorded in different widened lines, and of those bright forms which must appear on photographs taken with Fraunhofer lines that are weakened in intensity or transformed into bright lines in sun-spots, should provide data for the solution of important questions relative to the solar constitution.
But the spectroheliograph, though promising to supply an exceedingly powerful method of attack, is only one of many instruments required in a comprehensive investigation of solar phenomena. Powerful spectroscopes, equaling or surpassing in resolving power the largest instruments now employed in the physical laboratory, must be used simultaneously with the spectroheliograph in a study of the various vapors. In such a research the displacements of lines in various regions, corresponding to the effect of pressure or to the motion of the gases in the line of sight, would play a prominent part. This most precise quantitative work must lie accompanied by a systematic record, extending through many sun-spot periods, of the lines which are widened in sun-spots. Furthermore, there should be accurate measurements, with the bolometer or radiometer, of the heat radiated from