324 PROFESSOR L. BECKER ON THE DISTRIBUTION OF 



and A 6 and express the other unknowns as functions of them. The result of the 



solution is 



x = 140 +9-2 16 -22-2 A<< + 21'2 A/j 



y= 4-0 + 0-71 0-22- 4-68A/ 4 + 4-40A4 



*, = 0-84-0-28 0-08+ 3'01A* 4 - l'84A/ 5 



/ 2 = 0-80-0-18 0-06+ 2-28AA,- 1'20A* 5 



/ 8 = 0-78-0-01 0-04+ 1-66&4- 0'65A/ 5 



/ 4 = 0-80 + 1 A< 4 



< 5 = 0-85 1 A< 5 



The errors are mean errors. The mean error of an equation of unit weight is 

 0'014, as compared with the adopted value O'OIO. 



So far the time records have not been used (except in the calculations of the 

 differential quotients, which is merely a matter of convenience). I determine A 4 and 

 A? s from all the time records, introducing the condition that the values of t differ 

 from a mean value by accidental errors v. The equations are 



/- = 0-56 + 3-01 (A* 4 -A< S ) + 

 t 9 -v= 0-62 + 2-28 +1-08 



to-v = 0-77 + 1-66 +1-01 



<o-f = 0-80+1-00 +1-00 



/o-v = 0-85 +1-00 



The result is A 4 -A? 5 = +0-096-0'05 A 8 . The equations do not determine A< B 

 with any degree of accuracy. The unknowns then become 



x= 147 16 +2 A/j 



y = 4-26+ 0-22-0-05A4 



i, = 0-85+ 0-08 + 0-98A4 



/ 2 = 0-84 0-06+1 -07 A< 5 



/= 0-93+ 0-04 + 0-93A/5 



< 4 = 0-90 +0-95 A/.-, 



t b = 0-85 +1-00 A/.-, 



The value of A 8 is irrelevant for our purpose, it cannot be more than a fraction of 

 ft second and such a value changes x and y only by a small fraction of its error. 



We may change y by a small quantity 77 of the order of its error and x by 

 a corresponding quantity without altering appreciably the residuals. The equations 

 give = (1'880) r). I assume y = 4 and throw its error on *. The result is 



(A) z=12723, y=4-00. 



(6) Photographs Vb., VI. and VII. The photographs furnish material for the 

 intensity-curve from h = 110 to about 1700 (17 solar diameter). 



