( 473 ) 



furnished the data : 



0.015 8* = 0.006 



0.058 f^ ~ 0.022 



0.127 8^ = 0.051 



leading to the most probable values 8*=: 0.397; (7 =: 1.288. 



The deviations Obs.-Comp. have been given in the two last columns. 



Having found r^, we get y. and X from the light-intensities at the 



two minima. The values of these being 0.3433 and 0.6365, we 



obtain, for i = 90^, the two relations 



IX \ f y?l ^ 



= 0.3433 ; -11 = 0.6365, 



q 1 ^ y?' q y X 4- y 



whence : 



x = 0.6387; A = 0.3233. 

 Finally, at the moment at which the eclipse begins : 



The consideration of the asymmetry, shows that this must be 



the case shortly after 18'» {ntu = ± 20°55'). We therefore put : 



o a cos 21° 



^ = 1.6387 = l+x 



ƒ 1/1—8'' siV 21' 



from which : 



a =1.710. 



We thus have, as a first approximation, the following elements: 

 y. = 0.6387; A =0.3233; (7=1.288; a =1.710; e = ; i = 90° 

 and, as for the "epoch", we assume, that the central eclipse of E^ 

 by E\, coincides with the principal minimum of Argelander's curve. 



11. In the following table the 2"^^ column, headed C\, shows the 

 light-gi'ades of Argelander's curve for equal intervals before and 

 after the principal minimum ; the 7"^'^ column, headed 0^, similarly 

 shows the same element before and after the half period =: 6^^.455 

 {not therefore before and after the secondary minimum, which 

 Argelander places at 6"'. 375 from the principal minimum). The 

 columns Cs^ and Cs„ contain the light-grades, computed by the aid 

 of the elements given just now. 



12. As will be remarked, the deviations — Cs are in the main 

 negative before, positive after the two minima. We conclude that, 

 by shifting the theoretical light-curve in a negative direction with 

 regard to the time, we may obtain improved agreement. The excen- 



