undergoing changes and transformations well calculated to assist in the explanation of problems which the laboratory can not solve.

But to the philosopher and the student of nature as a whole, the sun finds its highest interest in its relationship to the great problem of stellar evolution. For the central body of our system is a star, resembling in the closest way many of the stars of the sidereal universe, but possessing the unique distinction of comparative proximity to the earth. Even in the most powerful telescopes, all the other stars appear as minute points of light, which every improvement in telescope construction tends to render more minute and microscopic, so great is the distance of these stars from the observer on the earth. We have no reason to believe that telescopes will ever be constructed so powerful as to magnify a stellar image into an actual disk. With our present knowledge we therefore may not expect that the great flames and other evidences of eruptive phenomena, which we believe from inference to be as characteristic of the stars as of the sun, will ever become visible. We must therefore depend for a knowledge of such phenomena upon the one star whose surface can be studied in detail. Armed with this knowledge, we may trace out with the spectroscope successive steps in the development of a star, from its origin in a nebula, on through the earlier stages typified by such stars as Sirius, to the condition attained in the sun. In this object we seem to observe the culmination of stellar life. For evidences of decay we must investigate the red stars, in which the radiation of heat throughout immense periods of time has resulted in cooling toward the point of final extinction. Thus we may untangle the great problem of stellar evolution, and at the same time build up a complete history of the sun, learning what it has been, what it is and what it will become.

Prior to the middle of the nineteenth century, at periods of total eclipse, when the dark body of the moon cuts off completely the bright light of the solar disk, red flames were observed at many points on the moon's circumference. At first their nature was so little understood that they were described by some observers as lunar mountains. But in 1868, through the use of the spectroscope, their true gaseous nature and their connection with the sun became known. It was found that immense masses of hydrogen and helium gas rise from a sea of flame (the chromosphere) which completely envelops the sun, and that these 'prominences' sometimes attain elevations of hundreds of thousands of miles.

The rarity and brief duration of total eclipses would have limited. greatly our knowledge of the prominences, had not a method been devised by which they can be observed on any clear day in spite of the glare of our atmosphere around the sun. The instrument which permits this result to be accomplished is the spectroscope, used in con

junction with the telescope. The principle of the method is simple and easily understood. The white light of the sky, when passed through a spectroscope, is drawn out into a long rainbow band, and thereby enormously reduced in intensity. The light of the prominences, on the contrary, is concentrated in the radiations characteristic of hydrogen and helium gas, and the great dispersing power of the spectroscope merely separates more and more widely the colored images which correspond to these radiations, without greatly reducing their intensity. With the spectroscope they therefore become visible, since their images are brighter than the highly dispersed background of skylight on which they lie.

Armed with this method, observers in various parts of the world have systematically observed the forms of the solar flames on every clear day, giving us a continuous record of these phenomena now extending back for more than thirty years. From a study of this record many conclusions regarding the nature of the flames and their bearing on the question of the solar constitution, have already been reached. But the process of observation is not only slow and painstaking; it is also subject to the errors and uncertainties that attend the hand delineation of every object, seen through a fluctuating atmosphere, under unfavorable conditions. It was principally in the hope of simplifying this process, and of rendering it more rapid and more accurate, that the spectroheliograph was devised by the writer in 1889.

The principle of this instrument is very simple. Its object is to build up on a photographic plate a picture of the solar flames, by recording side by side images of the bright spectral lines which characterize the luminous gases. To accomplish this an image of the sun is formed by the telescope on the slit of a spectroscope. The light of the sun, after transmission through the spectroscope, is spread out into a long band of color, crossed by lines representing the various elements. At points where the slit of the spectroscope extends out beyond the sun's edge across a gaseous prominence, the bright lines of hydrogen and helium may be seen extending from the base of the prominence to its highest point. If a series of images of such a line, corresponding to different positions of the slit on the prominence, were recorded side by side on a photographic plate, it is obvious that they would give a representation of the form of the prominence itself. To produce such an effect it is only necessary to cause the solar image to move at a uniform rate across the first slit of the spectroscope and then, with the aid of a second slit, which takes the place of the eyepiece of the spectroscope, to isolate one of the lines, permitting the light from this line, and from no other portion of the spectrum, to pass through the second slit to a photographic plate. If the plate is moved at the speed with which the solar image passes across the

first slit, an image of the prominence will be recorded upon the plate. The principle of the instrument thus lies in photographing the solar

flame through a narrow slit, from which all light is excluded except that which is characteristic of the prominence itself.

This method, when tried by the writer at the Harvard Observatory in 1890, proved unsuccessful. The lack of success was partly due to the fact that a line of hydrogen was employed. This line, though fairly suitable for the photography of prominences with the perfected spectroheliograph of the present day, was too faint for successful use

amidst the difficulties which surrounded the first experiments. Accordingly, when the work was resumed a year later at the Kenwood Observatory in Chicago, an attempt was first made, through a photographic investigation of the violet and ultra-violet regions of the prominence spectrum, to discover other lines better fitted for future experiments. In the extreme violet region, in the midst of two broad dark bands which form the most striking feature of the solar spectrum, two bright lines (H and K) were found which were attributed to the vapor of calcium. They had previously been seen visually in the prominences, but on account of the insensitiveness of the eye for light of this color, their true importance had hardly been realized. A careful study soon showed them to be present in every prominence observed, at eleva

tions above the solar surface equaling or exceeding those attained by hydrogen itself (Fig. 1, a). Their suitability for the purpose of prominence photography is due to several causes, among which may be mentioned their great brilliancy, their presence at the center of broad dark bands which greatly diminish the brightness of the sky spectrum, and the comparatively high sensitiveness of photographic plates for light of this color.

While fairly efficient from an optical point of view, the spectroheliograph of the preceding year had possessed many mechanical defects. In a new instrument, devised for use with the twelve-inch Kenwood telescope, these were overcome, and means of securing the necessary conditions of the experiment were provided. The first trials.

FIG. 3. ERUPTIVE PROMINENCE OF MARCH 25, 1895. a, 10h 34m. HEIGHT, 135,000 MILES. b, 10h 58m. HEIGHT, 281,000 MILES.

of the instrument, made in January, 1892, were entirely successful, and the chromosphere and prominences surrounding the sun's disk were easily and rapidly recorded (Fig. 2). The details of their structure were shown with the sharpness and precision characteristic of the best eclipse photographs. And the opportunity for making such records, previously limited to the brief duration, never exceeding seven minutes, of a total solar eclipse, was at once indefinitely extended. Thus it became possible to study photographically the slowly varying forms of the quiescent, cloud-like prominences, and, to particular advantage, the rapid changes of such a violent eruption as is illustrated in Fig. 3.

But even before this primary purpose of the work had been accomplished, the possibility of making another and much more important application of the instrument had presented itself. A photographic