142 THE ROYAL SOCIETY OF CANADA 



obtained by dividing the stars into three main groups according to 

 the method of measurement and the number and quality of the 

 spectral lines. All the stars of type FO to M, with the exception of a 

 few fuzzy lined F's, have been measured on the spectro-comparator 

 and form the first group with probable errors of the mean ranging 

 between ±0.1 and about ±1.0 and of a single plate between ±0.2 

 and ±2.5 km. per second, with average errors of ±0.5 and ±1.2 

 respectively. The second group includes the A and B type stars 

 with fairly sharp and numerous lines, measured on micrometer engines, 

 the probable errors ranging from about ±0.5 to ±1.5 for the mean 

 velocity and from ±1.2 to ±3.5 for the single plate. The third 

 group includes those stars, mostly of the A type in which the few 

 lines present are broad, diffuse and frequently lacking in contrast. 

 In many of these only broad H7 and H5 were measurable and the 

 accidental errors are high, ranging from about ±1.0 to ±4.0 for the 

 mean and ±2.5 to ±10.0 for the single plate. To compensate 

 partly for the unavoidabJy high accidental error of this group, a 

 larger number of plates, up to 8 or 10, were made of many of these 

 stars. For single prism dispersion the accuracy of these velocities 

 may be considered quite satisfactory and the number of plates ob- 

 tained increases the confidence in the reliability of the results. 



The complete publication will contain, besides descriptive matter, 

 tables of the individual velocities of 3,493 plates of 572 stars. These 

 velocities are summarized in tables of the mean velocities of 537 

 constant velocity stars and of the estimated velocities of the centre 

 of mass of 35 spectroscopic binaries. In addition a table of the 

 "gamma" velocities of 22 spectroscopic binaries, whose orbits have 

 been determined here, is given. 



