are 64 and 16 respectively. The difference between the two values is 
highly significant. This means that the aperture changes the image 
coordinates of the fiducial marks. This will be discussed in relation 
to the principal point. 
Principal Point 
The results of the calibrations are given in Table 1. The variation 
between pictures with the same aperture is not significant for any 
aperture. The variation between the aperture is significant for the 
x-coordinate but not for the y-coordinate. The correlation coefficient 
between x and 7 is 0. 08 and therefore the combined effect is not 
significant, i. e. the principal point does not vary with the aperture. 
The position of the principal point is given in a coordinate system 
defined by the fiducial marks. If we want to have the fiducial marks 
without errors, the standard errors of the fiducial marks have to be 
propagated to the coordinates of the principal point. It is also possible 
to let the principal point define the fiducial coordinate system and then 
the standard errors of the principal point have to be propagated to the 
fiducial marks. In this calibration the standard errors of the princip 
al point will dominate the resulting standard errors whichever 
method is used. 
Camera Constant 
The camera constant refers to a radial distortion that is zero for 
radius 8 mm. The results are given in Table 2. The variation bet 
ween pictures with the same aperture is not significant. For aperture 
f: 5. 6, the variation is almost significant. The variation between 
apertures is highly significant. 
Radial Distortion 
The radial distortion curves and their standard errors are shown in 
Fig. 1. The difference between the curves are not significant for 
radii larger than 4 mm. 
Standard Errors of Image Coordinates 
In the adjustment of the observations the image coordinates are 
given unit weight. The accuracy of the image coordinates is then ex 
pressed by the standard error of unit weight (s Q ) in the adjustment. 
The homogenity of the values s^ can be tested by Bartlett's test. The 
t^st is, in this case, rather weak, because of the small number of 
s 0 : s, but they can be regarded as homogeneous for all pictures 
having the same apertures. The variation between apertures is 
highly significant. The values are given in Table 3.