( 719 ) 



If now we reduce all these equations (o the same weight, so that 

 the p. e. of theii- right-hand members becomes ± O.iO, we find, if 

 also the signs of 1 are reversed: 



Res. 

 lies. SouiLL. 



+ 33.3rfK' f 3.8ffr, -f O.lrfr, -f 5.0rfr, +0.2r, = +0.12^ -I-.02 -.78 



(C) 



The linally adopted corrections are: 

 (iyj = + 0.005 ± .0075 



(fv, - + 0.010 ± .030 rfr, = ± .050 



ÖV, — — 0.020 ± .020 r, = zt 0.25 



The corresponding values of the masses are: 

 Jb' = 0.0214 180 ± .0001543 {b = 1 for (/ = 39".0) 



= 0.0000 0000 518169 db 3975 (astroiiuraical units) 



j/tj = O.fiOOO 260 ± .0000 012 



m, = (1.0000 231 ± 11 



7rtj = 0000 804 ± 16 



m, = 0.' 000 424 751 ± .0000 106 



Substituting these corrections, there remain the residuals stated 

 above. If Souillakt's masses are substituted there remain the resi- 

 duals given in the last column. 



The equations II and III are contradicting each other: II demands 

 a negative value (fv^, III a positive value. On account of the bad 

 agreement of the different determinations of d\ I have assigned a 

 very small weight to the equation III. It is to be noticed that the 

 large negative correction dv^ could have been partly avoided by 

 assuming a large positive \alue of i\, e. g. i\ = + ^•^- Even then, 

 however, it would not be possible to bring about a satisfactory 

 agreement of II and 111 without spoiling the re|)resenlation of I 

 and IV. 



The probable errors stated for the corrections ex' and dr,- as well 

 as the values of these corrections themselves, depend largely on judg- 



