218 ROYAL SOCIETY OF CANADA 



It was found, however, that lower probable errors were obtained 

 where the three besthnes X 4481, 4472, 4341 only were used than when 

 more or all of the measured lines were discussed. Consequently in the 

 later plates only the three lines mentioned above have been measured and 

 in general four comparison lines thus considerably lightening the labor. 

 Considerable difference in the quality of the negatives for measurement, 

 even when taken under so far as could be judged identical conditions, has 

 been noticed. This difference seems to lie principally in the character 

 of the lines themselves. They are sometimes sharply defined and sym- 

 metrical, at others not so sharp and appairently stronger at one side. 

 Sometimes also the contrast between them and the continuous spectrum 

 appears considerably diminished. These changes seem almost too marked 

 to be due entirely to instrumental or photographic effects, and one would 

 be inclined to attribute part at any rate to changes in the spectrum. Xo 

 evidence can be found, however, of any dependence of this quality of the 

 lines upon the phase in the orbit. 



In the measurements the lines were weighted according to their ap- 

 parent quality and the weighted mean velocity used. In combining the 

 separate plates on each night they were also weighted, partly according 

 to their quality, and partly according to the internal agreement of the 

 measures and finally the resultant mean velocity for the night was 

 similarly weighted for use in the grouping and discussion. 



'The individual measures with a summary will be found in my report 

 to the Chief Astronomer for 1908-190!), but it has not seemed necessary 

 to give them here. The mean velocities for each night were obtained by 

 taking the weighted means of the plate velocities, the weights being as- 

 signed, as before stated, partly on the basis of apparent quality, partly 

 according to the internal agreement of the measures. In the following 

 table of mean velocities are given various data concerning the observa- 

 tions of each night as the date, Julian Date, velocity, phase, the number 

 of plates, the dispersion used, the weight assigned and finally the residual 

 obtained by scaling from the curve. 



