If the normal equation system as given with formula (51) is directly 
inverted, the computation of the corresponding mean errors for both the orien- 
tation elements and the coordinates of the model does not cause any difficulty. 
Correspondingly, the mean errors of the unknown parameters associated with the 
reduced normal equation system in the method of partitioning (x, in formula 
(28) or x, in formula (29), respectively) are directly obtained using the 
diagonal terms of the inverse of the corresponding normal equation matrix, To 
prove this, we write the matrix of the coefficients of the normal equation 
system (35) with a notation corresponding to formula (24) as: 
  
  
  
(60) 
C 
The corresponding inverse matrix shall be denoted by: 
Q | Qı2 ; 
T (61) 
Quo Qo» 
0 
According to the definition of the process of inversion, with respect to the 
notation used in formulas (60) and (61) follows: F 
Au À Qu Q 
Ww Pe ti da} 5 | 
T T T 
Al Az Qi Qoo (0 I | (62) a 
e 
De 
42