MOUNT WILSON OBSERVATORY. 199 



comprises stars selected for observation during recent years for a wide variety 

 of reasons, occasionally on the basis of apparent magnitude, but more often 

 because of large or small proper motion, variability, binary character, or some 

 similar characteristic. The composition of the list is related closely to the 

 work upon the spectroscopic determinations of absolute magnitude and 

 parallax, many stars being observed primarily for spectral type and only 

 incidentally for radial velocity. As a consequence, the list is by no means 

 a representative one for stars in general but includes a large excess of giant and 

 dwarf stars. 



A sufficient number of stars has been observed in common with the Lick 

 and the Dominion Astrophysical Observatories to make it possible to compare 

 the results for systematic differences. In the case of 109 stars of types F to 

 M observed at the Lick Observatory, the Mount Wilson systematic differ- 

 ence is —0.12 km., the agreement being excellent. Similarly for 83 stars 

 observed in common with the Dominion Astrophysical Observatory, the 

 difference is +0.77 km. In view of the low dispersion used in the spectro- 

 graphs at Victoria and Mount Wilson, this value may be regarded as satis- 

 factory, although the rather large systematic difference found for stars of 

 type M requires further investigation. 



During the past four years Merrill has obtained observations of the radial 

 velocities of 113 long-period variable stars. With the addition of those for 

 which velocities were known previously, a total of 133 stars of types M and 

 S is now available for which radial velocities have been derived from the 

 bright lines. Velocities from the absorption lines have also been measured 

 for 47 of these stars. The relative displacements of the bright lines are 

 found on the average to increase with advancing type, increasing period, and 

 magnitude range. The correlation with period has been used to establish 

 an empirical correction to be applied to the bright-line velocities to reduce 

 them to a dark-line basis. This is essential, since the displacements of the 

 dark lines rather than those of the bright lines appear to correspond to the 

 true radial velocities. The investigation has necessitated the determination 

 of the wave-lengths of numerous bright and dark lines in these spectra. 



Studies of the apparent solar motion with respect to these 133 variables 

 show that the speed of the sun is almost three times that usually found for 

 K and M stars, but that the position of the apex is nearly the same. The 

 following values are representative: Ao = 281°; Do=+34°; 1^0 = 53 km.; 

 iC=+l km.; arithmetic mean residual, 31 km.; 68 stars with residuals less 

 than 25 km. give Fo = 48 km., and 65 with larger residuals, Fo = 65 km. This 

 increase in Vq furnishes an excellent illustration of the well-known velocity- 

 asymmetry of stars of high speed. The average residual radial velocity is 

 found to decrease with advancing spectral type and increasing period. Very 

 high velocities are confined largely to stars of types M2e and M5e and to 

 stars having periods in the neighborhood of 200 days. These facts are diffi- 

 cult to explain on the basis of any hypothesis of the origin and evolution 

 of red stars which has yet been proposed. 



Several stars which were observed over a considerable part of the light 

 curve show a slight variation in the apparent velocity from the emission 

 lines. The velocities appear to have algebraically low velocities for a month 

 or two after ma.ximum of light. 



