EE ERR REIR ts 
  
  
  
i 
  
   
pancies are small. Therefore the method is normally to be used iterati- 
vely. However, experience has shown that almost all models show mean 
square parallax values of around 0.01 mm in the pictures after two or 
three iterations. This figure can be slightly decreased by a final adjust- 
ment of the by-setting as an average of the measured y-parallaxes in all 
9 points. The fastest results are obtained through a combination of a 
simple empirical method and the numerical one in one or two steps. 
Further it was found that the standard error of the y-parallax measure- 
ments according to the adjustment procedure normally does not exeed 
0.006 mm in the pictures.*) If corrections are numerically computed to 
the elements of the reciprocal orientation and functions thereof, this 
figure (0.006 mm) can be used as the standard error of the measure- 
ments. 
The form (Fig. 1) has been designed so that it can be used in all 
cases of orientations, viz. base in or out, independent or dependent mo- 
del, and succeding picture to the right or left. Correct signs are auto- 
matically obtained for all these alternatives. On each sheet there is space 
for three calculations of corrections, and in the last two there is also 
room for calculations of the standard error fu of the y-parallax measure- 
ments. The form also includes the formula for computation of mean 
: / [by] [by]? 
square values of the y-parallax, mg = V/—— — — 
9 81 
adjustment as an average of the measured y-parallaxes. A comparison 
between u and mg will then show wheter still an iteration is required. 
The form also constitutes a valuable record of the procedure and the 
degree of accuracy obtained. 
For a comparison with empirical methods reference is made to Schwi- 
defsky [3]. On page 204 he quotes Bachmann, who has found e. g. 
= / 3 h? ; 
my = + V 5 ari "0 
  
, after a by- 
mo is defined as the standard of the measurements, but it is not clear 
how this coefficient is or schould be determined. 
From [1] the standard error for the same element of orientation may 
be calculated as 
/3. .h? 
m,, — fo motio 6 points of orientation and 
e A 4 d*í(^-^ 
from [2] as 
— + — te ato 
me, x V2 u o for 9 points. 
- 
*) Occasionally values of 0.003 mm are obtained.