( 248 ) 



tation for the atmospheric pressure, the temperature, etc. on January 14, 

 1904. Hence on Jan. 14, 1904 c'^, c^, c^\ c^" diffei- from zero and 

 in the formuhi I had to retain the term for the end. 



nP S'^ iP (S—f) 



In inteQTals such as I -— ; — — dt 1 lirst put all errors equal 



® J dt" de ^ 



a 



to zero and computed the known terms of tlie four equations accord- 

 ing to formulae of the form : 



a 



Hy solvin.ii' tliesc equations, 1 have ohlainrd |Mvlimiiiarv \alues 

 for ./', y, z and //, and in tlu^ way to be explained iu § 7, also for 

 the most probable errors /' ; tVom the mean dailv rates computed 

 with these preliminarv \alues, using the formula 



Q,^ = Q - .,■ Qi' - li Q^ -zQy - n Qi' - -^^^^ 



n 



\ luiNC derived as a second approximalion the corrections ,c, //, z 

 and II to these preliminary values. The influence of those corrections 

 on the most probable errors was of little importance. 



r/d" sy 



Integrals such as I I — — I dt have been computed twice, first 

 according to the formula for \/, I ~~J^ ~^ '^^ 



\m) 



<// = [.,/] 4- ^Q„{c,,-~c,) 



and secondly according to the formula, deduced in § 2 



) '^^ = -5- -^ " (^7 + ^'1 ^'- + *^'- ) 



ƒ( 



The following are the 4 equations expressed in numbers : 

 5460 X — 252 ij— 14 c -[- 59 u = -f 21 664 — 145 



— 252 ,1' + 952// + 512 ^— ^u=— 1757— 88 



— 14 ,v + 51 2 // + 571 z - 22 u = + 383 + 87 

 _|_ 59.6— 8//— 22c + 101z^ = — 101^ 21 



