( 54 ) 



tiou of such a result was computed ; bv Albrecht himself the 

 comparison of the observations with his curve had already been 

 made. Moreover for every observatory Albrecht's residuals as well 

 as mine were united to half yearly means, that is to say means of 

 these residuals were taken for each summer- and each winter-half year, 

 and finally from these, mean deviations for those half-yearly periods 

 were computed. ') 



It would take too much space to communicate these comparisons in 

 extenso, so I shall restrict myself to a few results derived from them. 



If we call Aa <ind Ab the mean residuals of the single results 

 of observation as found by Albrecht and by myself, computed for 

 each observatory ; A'a and A'b the corresponding quantities for the 

 half-yearly periods, I find, taking the means of the 27 values of 

 each of these quantities : 



A^ (mean) = ± 0".O61 A'^ (mean) = i 0".033 



Lb (mean) = ± .071 A'^ (mean) = ± .048 



Among the 27 series A^ is smaller than A^ in 5 cases. (Pulkowa 

 2»"^ sei'ies, Berlin \^^ s., Cape of Good Hope, Potsdam 2"^ s. and 

 Lyons) and larger in 22 cases; for the A' the corresponding numbers 

 of cases are also 5 and 22. 



If I use only the 18 series with the smallest mean deviations 

 from Albrecht. I find 



A^ (mean) = ± 0".051 A'^ (mean) — dz 0".029 



A^ (mean) = ± .061 A'^ (mean) = ± .041 



The fact that the Lb are in general larger than the A^ may 

 be attributed to real deviations from my formula aud to systematic 

 errors of the observations, which for a part will have been trans- 

 mitted to the curve of Albrecht. As mean values for the sum of 

 those influences (deviation from my formula and partial influence of 

 the systematic errors), we find for all observatories together : 



27 '^ 27 



and for the eighteen most accurate series only : 



'^ 18 ~ 18 



So it appears: P that the combined influence of both circum- 

 stances is very appreciable; 2° that influences ai'e at work which 



*) Means depending oaly ou a single monthly result were neglected. 



