( 418 ) 



The final result of llie foreuoiii^' iin estiuatioii is therefore that 

 the observations eonfiiiii the existence of the inequalities III and IV, 

 hut that they seem to eontradict the existence of the term I. 



J6. I then niaile a new (leterniinalion ot' the principal term of 

 the eni|»irical inecpiality, after first havijiu' corrected the h and /• 

 for the ine(jualities III and \\\ for which I accepted the theoretical 

 values. 



From the h and /■ for each year corrected thus, from which 

 moreoxer the constants //^. =rz -|- ()".55 and /•,. =r -[- ()".26 were sidi- 

 tracted ^), 1 have first derived the values of A. They follow hei'e : 



If to these values xve assign the same weights as before, we obtain 

 the followinü: normal values : 



1) These values of the constants were found as a first approximation after the 

 inequality HI had been accounted for. The differences between them and those 

 wliich were finally found, +0".43 and +0".17 are unimportant in comparison 

 with the tireat systematic discordances whicli still remain. 



