Til K OKI! IT () I I K ANUS. 



\ 



A simple glance at the course of the residuals shows (1) that their probable 

 value is considerably greater than the probable error attributed to the equations 

 of condition, being more nearly 0".7 than 0".5, and yet larger in the later years; 

 (2) that during certain periods they are of a systematic character. During the 

 years Ills to 17">3 the observations show a decided positive correction to the 

 theory of a magnitude greater than we can consider probable, amounting to about 

 one third of a second of time in the mean of Bradley's two observations of 1748 

 and \l')'-\. About 1800 the correction becomes negative, and so continues for 

 '() \. ;ns with an average value of about 1". In 1821 it suddenly becomes 

 positive, and so continues until 1833. From this year forward the residuals are 

 not systematic in character. 



In order to show clearly the general course of the outstanding corrections, they 

 have bedi divided into groups, generally including about five years each. The 

 mean outstanding correction for each group, taken with respect to the weights 

 indicated by the factors f, is as follows. In the column e is shown what the 

 probable error of the residual should be if the weights assigned to the several 

 equations were strictly correct, and no systematic errors were present either in 

 theory or observation. 



