red 
n 
Vb 
S 
  
  
dx, + dx, - ax. - ax, - dy, + dy, - dy 5 + dy, (17) 
P 
M 
9 = - dx, + dx, - dx * dx, + dy, + dy, - dy; - dy, (18) 
x = image ~ “projector (19) 
dy s “image ~ Yprojector (20) 
bi - [axax] + [ayay] - [a0 % + (ay]?) - àa + k2) (21) 
From [vv] the standard error of one observation is found as 
p= i (22) 
A number of pictures were empirically centered with a 
normal accuracy and the discrepancies between the fiducial 
marks in the image and those of the projector were measured 
in the instrument (the Autograph A7). To get an idea of the 
accuracy of a normal centering the corrections and standard 
errors are compuied accordins to equations (13) - (22). See 
table 1l. 
The size of the standard error is particularly interes- 
ting, since it is also influenced by the systematic discre- 
pancies between the two sets of fiducial marks, besides by 
the measuring errors. The standard error was quite large in 
the investigated cases and is evidently due to iregularity 
of form of the sets of fiducial marks.