[Edward C. Pickering: American Journal of Science, 3rd Ser., vol. 39, pp. 46-47, 1890]
In the Third Annual Report of the Henry Draper Memorial, attention is called to the fact that the K line in the spectrum of Zeta Ursae Majoris occasionally appears double. The spectrum of this star has been photographed at the Harvard College Observatory on seventy nights and a careful study of the results has been made by Miss. A. C. Maury, a niece of Dr. Draper. The K line is clearly seen to be double in the photographs taken on March 29, 1887, on May 17, 1889 and on August 27 and 28, 1889. On many other dates the line appeared hazy, as if the components were slightly separated, while at other times the line appears to be well defined and single. An examination of all the plates leads to the belief that the line is double at intervals of 52 days, beginning March 27, 1887, and that for several days before and after these dates it presents a hazy appearance. The doubling of the line was predicted for October 18, 1889, but only partially verified. The line appeared hazy or slightly widened on several plates but was not certainly doubled. The star was however low and only three prisms could be used, while the usual number was four. The predicted times at which the line should be again double are on December 9, 1889 and on January 30, 1890. The hydrogen lines of Zeta Ursae Majoris are so broad that it is difficult to decide whether they are also separated into two or not. They appear, however, to be broader when the K line is double than when it is single. The other lines in the spectrum are much fainter, and although well shown when the K line is clearly defined, are seen with difficulty when it is hazy. Several of them are certainly double when the K line is double. Measures of these plates gave a mean separation of 0.246 millionths of a millimeter for a line whose wave-length is 448.1, when the separation of the K line, whose wave-length is 393.7, was 0.199. The only satisfactory explanation of this phenomenon as yet proposed is that the brighter component of this star is itself a double star having components nearly equal in brightness and too close to have been separated as yet visually. Also that the time of revolution of the system is 104 days. When one component is approaching the earth all the lines in its spectrum will be moved towards the blue end, while all the lines in the spectrum of the other component will be moved by an equal amount in the opposite direction if their masses are equal. Each line will thus be separated into two. When the motion becomes perpendicular to the line of sight the spectral lines recover their true wave-length and become single. An idea of the actual dimensions of the system may be derived from the measures given above. The relative velocity as derived from the K line will be 0.199 divided by its wave-length 393.7 and multiplied by the velocity of light 186,000, which is equal to 94 miles a second. A similar calculation for the line whose wave-length is 448.1 gives 102 miles per second. Since the plates were probably not taken at the exact time of maximum velocity these values should be somewhat increased. We may however assume this velocity to be about one hundred miles per second. If the orbit is circular and its plane passes through the sun, the distance traveled by one component of the star regarding the other as fixed would be 900 million miles, and the distance apart of the two components would be 143 million miles, or about that of Mars and the sun. The combined mass would be about forty times that of the sun to give the required period. In other words, if two stars each having a mass twenty times that of the sun revolved around each other at a distance equal to that of the sun and Mars, the observed phenomenon of the periodic doubling of the lines would occur. If the orbit was inclined to the line of sight its dimensions and the corresponding masses would be increased. An ellipticity of the orbit would be indicated by variations in the amount of the separation of the lines, which will be considered hereafter. The angular distance between the components is probably too small to be detected by direct observation. The greatest separation may be about 1.5 times the annual parallax. Some other stars indicate a similar peculiarity of spectrum, but in no case is this as yet established.
Addendum, Dec. 17
The predicted doubling of the lines of Zeta Ursae Majoris on December 8th was confirmed on that day by each of three photographs. Two more stars have been found showing a similar periodicity: Beta Aurigae and b Ophiuchi (H.P. 1100 and 2909).
Harvard College Observatory, Cambridge, U.S., Nov. 12, 1889.
* Read at the Philadelphia meeting of the Nat. Acad. of Sciences, Nov. 13, 1889.