parison between different scales these errors are of no im- 
portance, because they are systematic and thus are corrected. 
Table 4 
Errors in <h (in %) caused by incorrect base due to tilt. 
  
Divergency 9 Convergency y 
  
0.5* — 1.5 9 + 16% 
1e — 3.0 % + 32 % 
2c — 5.9 9; 457659 
3¢ = 86 % +101 % 
2. 15. Effect of tilt on horizontal parallax differences ( Apt). 
With the camera axis not perfectly vertical the measurement 
of height differences will be incorrect, although the flying altitude 
is perfectly known and the base value is corrected, because tilt 
affects the parallax differences (Ap!) measured. In literature 
there does not seem to have been made any investigations on the 
magnitude of these errors except general statements saying that 
the errors will be of little importance for height difference deter- 
mination of adjacent terrain details. (For instance tree height 
measurements). To give an idea of the errors which can be ex- 
pected in this respect investigations have been carried out for 
some different cases. 
By means of the complete formula for the correlation between 
ground coordinates and photograph coordinates (Hallert): 
xl *[x (cos @ cos X — sin @ sin 0 sin x) — y cos @W sin © - h (sin 9 cos % + cos P sin wsin X | 
x sin P cos  — y sin € + h cos 9 cos 9 
x'= photograph coordinates, x = ground coordinates 
the parallax difference was calculated for a number of tilt angles 
and for height differences of 20 and 30 m respectively. The 
Apt thus calculated was compared with the Ap! obtained for 
truly vertical photographs. The error in parallax difference 
A h … : 
multiplied by pi gives the error in Ah. 
The magnitude of the errors is dependent on the location in 
the model of the measured details. 
10