I'.LUE-VIOLET LIGHT IX TMK SOLAR CORONA ON AUGUST 30, 1005. 325 



Again two of the Al remain indeterminate. I take A 7 = 0, which is permi&sible, 

 as any reasonable error has a small effect on x and y, see (c), and express the 

 unknowns as functions of AJ&. The result is 



*- 140 -23 46 +1-4 A/s 

 y- 4-00- 0-29 0-32-t-05A/ 4 

 / = 11-0 + 1-37 + 0-60 + 5-45A/4 



The error of an equation of unit weight is 0'021. 



The time records (except t^+t. = 100'88) have not yet been used. Af s can be 

 determined from the last equation. 



The upper limit of J [see 6 (a)] is 11 -84-0-1 = 117 and it gives Af 5 = -0'12 

 and the lower limit of t t is H'84 I'O = 10'8, which gives A/ 5 = 0*29, both with an 

 error of O'l 1. The large value of A* 6 belonging to the lower limit of t t is out of the 

 question, because the pendulum of the contact clock could not possibly have been 

 placed so much out of beat. Nevertheless, I maintain lx>th values, 



/- 11-7 / = 10-8 



z = 117 46 z = 117 46 



y - 3-84 0-34 y= 4-01 0-34 



/i = 0-73 /j = 0-56 



I again reduce x to y = 4'0. The normal equation gives = (2' 107) rj, hence 



(B) / - n-7 / = 10-8 



=137 63 t = 116 63 



y = 4-00 y = 4-00 



^ = 0-73 / a = 0-56 



The second result is not possible, as already mentioned. 



(c) All the Photograph I. to VII. In this solution I have included the unknown 

 Aa in order to see what effect the error of a hs on x and y. Five corrections AF 

 can be found or five of the A, leaving two, say AJ 5 and A* 7 , besides Aa, indeterminate. 

 The result is : 



t = 140 -2 19 -7 A/i + 0-5 -V 7 - 20 \* 

 y= 4-00-0-15 0-14-1-1 A/ 5 + 009 A/ T - 4 Aa 

 ^ /, = 0-84 + 0-21 0-08+1-68A/5-0-006A/7+ 2-5 Aa 

 / 2 = 0-80 + 0-19 0-06 +1-46 A/ & -0-004 A/ 7 + 2'0 Aa 

 /, = 0-78 + 0-27 0-05+ 1-30 A/ 4 -0-003 A/ 7 + 1-4 Aa 

 <4 = 0-80 + 0-17 0-04 + l-16A/ s -0-002A/ 7 + 0-8 Aa 

 < - 0-85 +1 A/ 5 



11-00+1-17+ 0-37 + 5-5 A/i-0-063A/;+ 0'25Aa 

 /? = 100-88-^ +1 A/ 7 



