( 381 ) 

 Epoch W/ujht Xo—^^c X'o—Xj 



1862.5 'A — 18 () 



That the differences, even those with the formula that is made to 

 represent the observations as well as possible, are not altogetlier 

 accidental, may be seen from tlie great numbci- of pei'manencies of 

 sign. Yet I liold tliat we are entitled to tlie conclusion that an annual 

 variation of .V of 1 9^.35 better represents reality than one of 20^65. 



Having thus derived a formula for .V representing as well as 

 possible the results at my disposal, I had still to correct the adopted 

 values of the coefficient a and of lie ^and kr. 



To this end I compared the obserxed values of h and k with the 

 formulae 



/t = -\- 0".45 — 1".50 sin [302°.4 + 19°.35 U — 1876.0)] 

 /■ = + 0".26 -f 1.".50 co.s [302°.4 -f 19^35 {t — 1876.0)] 



and formed the outstanding resi(hials Obs. — Com|). These residuals 

 which for shoi'tness arc not give]i here, were divided into 4 groups 

 according to the 4 (piadrants of ^V, and for eacli of these groups 

 mean values were formed whifh follow here: 



