Stellar Due to the fact that gratings spread the light into a number of or ders of spectra, prisms are almost exclusively used in stellar investigations. A prism may be combined with the telescope, either with or without a slit. If without a slit the prism is placed outside the objective. In such a com bination the dispersion is usually not very great. At Harvard College Observatory, where the best work with this method has been done, Pickering combines a prism of 15° angle with an objective of 11 inches aperture. In this manner are obtained spectra of a large number of stars at one time. To increase the width of the photographed spectra, the prism is mounted with its edge parallel to the celestial equator, and the clock-work driving the telescope is rated so as to run a little slow, or fast. To in crease the dispersion, a second or third prism may be added. With the objective prism, it has so far been impossible to obtain radial ve locities with the accuracy that is attained with a slit spectrograph, though Pickering has superimposed some very satisfactory additional lines in the stellar spectra by means•of the ab sorption of neodymium. When a slit is used, it is placed at the focus of the objective. After passing through the slit the light passes through the collimator lens, from which it emerges parallel. The light is refracted by the prism, or prisms, and is brought to a focus by the camera objective, so that the spectrum may be photographed or viewed with an eyepiece.
Photography is almost exclusively used in mod ern work. One, two or three prisms may be used to obtain dispersion. Needless to say, in crease in the dispersion adds to the time of ex posure necessary to obtain a photograph.
Modern stellar spectroscopes are best repre sented by the Mills spectrograph of the Lick Observatory (Campbell, Astrophysical Journal, VIII, 123, 1898), by the Bruce spectrograph of the Yerkes Observatory (Frost, Astrophysical Journal, XV, 1, 1902), and that of the Astro physical Observatory of Potsdam (Hartmann, Zeitsch. fiir Instrum., December 1901). The spectrographs are similar in having the slit placed at the focus of the great telescope, and a dispersion of three prisms giving a total de viation of 180°. By means of a guiding eye piece, it is possible to keep the star's light cen trally on the slit during the exposure. Since this exposure may last for four or five hours, it is necessary, in order to have perfect defini tion, to keep the temperature constant. This is accomplished by means of an automatic tem perature control which will keep the prisms of the spectrograph within 0.1° C. during an ex posure, when outside in the dome the tempera ture may change by several degrees. A stellar spectrum is photographed alongside a compari son spectrum in order to determine wave lengths more accurately, and to give measures of the motion in the line of sight, the most im portant work of stellar spectroscopy. The spec
trograms are most readily reduced by the Hart mann-Cornu formula: = + 5-50' where 24, c and s. are constants and s is the scale reading of the line whose wave-length is desired.
It needed little investigation of the stars to show that difference in color of the stars cor responded to differences in the character of their spectra. One of the best known classifi cations is that of Secchi, who divided the stars into the following types: Type White or blue stars, the spectrum characterized by the breadth and intensity of the hydrogen lines with metallic lines very faint. This type includes more than half the stars.
Type stars like our sun, with spectra resembling that of the sun very closely, consisting of a great number of fine dark lines.
Type III.— Red and orange stars, including most of the variables. The spectrum is crossed by numerous dark bands or flutings, which are sharply defined on the blue side and shade off toward the red. a Orionis, Antares and 0 Ceti are good examples.
Type IV.— Deep reddish stars, all faint. The spectrum resembles that of Type III, but the flutings are reversed in direction, being sharply defined on the red side. 152 Schjellerup is the best example.
Pickering has added a fifth type to include many stars having bright lines in their spectra, and the planetary nebulae.
Other satisfactory classifications are those of Vogel and Lockyer. The classification most used at the present time is the one devised at Harvard College Observatory. In this, not only are there the main classes, but also sub divisions of each class as, BO, Bl, B2, B3, . . . B9. The main classes are: Class 0.— Stars with bright bands. Wolf Rayet stars. Typical stars Y Velorum, Puppis.
Class B.—Helium stars, represented by 6 and e Orionis.
Class A.— Hydrogen stars, represented by Syrius and Vega, and characterized by the great intensity of the hydrogen lines.
Class F.— Calcium stars, in which H and K have the greatest intensity; represented by 6 Aquila and a Carina.
Class G.— Solar type, represented by the sun and Capella.
Class K.— Represented by Arcturus. There is a decrease in intensity at the violet end of the spectrum.
Class M.— Represented by B Andromeda, a Orionis, a Hercules and Mira Ceti. Absorp tion bands are found in the blue-green.
Class N.—Similar to Secchi's type IV.
Miss Cannon of Harvard College Observa tory has completed the classification of 250,000 stars according to the above system. This clas sification seems to represent successive steps in the evolution of the stars.