( 556 ) 

 TABLE V. 



The first problem is identical with the calculation of the prob- 

 ability of an event n -+ h, when d and b follow the normal law and 

 are independent of each other. 



This problem of the superposition of two laws of errors has been 

 already treated by Bessel ^) and subsequently d'Ocagne '') gave a 

 general solution for the superposition of several groups of errors. 



It appears that, if H be the factor of stability of the secular and 

 /<! that of the interior variability, the resulting deviations also follow 

 the normal law, the new factor h. being determined by. 



/, = ^^^ 



From the values of h given in Table IV and those of H calcu- 

 lated from monthly means we can, therefore, deduce that of h^ -. 



1) Untersuchungeu iiber ilie Walirschcinlii^likfil der BeubuclituDgsfehlcr. Astr. 

 Naclu'. XV, 1838. 



-) Sui' la cumposiliuii des lui.s d'eri'cui's do sitiuUiun d'uii point. G H. Acad, 

 sc. CXYlll, 1894. 



