3. The Fertsch effect has been studied and is estimated as a systema- 
tic height error in the contour lines when they are about parallell to the 
stereo base. The height errors are of the magnitude 0.0, 0.1, 0.1 and 0.2 
metres in the four different tests. 
Theoretic study of the height determination 
Decause of relative and absolute orientation there are systematic errors 
in the heights of the stereomodel. The magnitude of the systematic 
component can be estimated by error propagation from standard errors 
of image coordinates and parallaxes and of known geodetic heights. 
Tests indicate standard errors of unit weight of image coordinates and 
parallaxes to be 6 um and 8 yum in estar-film (black-and-white) and 
acetate-film (colour) respectively. From Table 5 we obtain a standard er- 
ror of unit weight in the absolute orientation to be 0.027 metres. 
The errors of the heights also contain an accidental component that 
varies within the stereomodel. This accidental component depends on 
the standard errors of the image coordinates and on the type of terrain 
that is measured. The latter part is estimated to 
0.040 metres for rocks 
0.070 metres for open fields 
0.100 metres for terrain with forest. 
Based on these assumptions one can evaluate the error propagation In- 
cluding the compensating effects (the same technique as in Hallert, 
1956) and the standard error of the elevation as estimated by the follow- 
ing formula 
G2 — K2s2 + 52 + K2s? F s2 
H rel > abs “phot terr 
where the first two terms yield a systematic component and the others an 
accidental one for each model. K is the scale factor of the negative. 
Assuming for example 600 metres flying altitude, black-and-white 
polyester film and mainly open fields we obtain 
Su — 0.079 metres 
This value can be compared with s — 0.103 metres in Table 5. The 
difference depends on the uncertainties of the geodetic control measure 
ment. Its standard error can be estimated by quadratic subtraction 
S — 0.060 metres 
geod