+/6 -34 "64 1/0 -3 FG -ó 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
*£3 f - £6 
-6 
+/ - 
+08 -20 -43 
+/2 "Jo -64 NI 
E 
5 
Oo 04 -05 22 0 02 
Qo -Qa -Ja 
-g 
~/ 
-06 +09 -Q, 
£g E -Q 0 
-/ 
+/ 
‚46 *££| */? „9 
£^ ds * £e 40 4 22 : 
Figure 7. 
Errors in height differences of 20 resp. 30 m, 
P = 0.607, o — — 0.705, z = 0.50, h = 4 000 m, c — 0.15 m. 
It is assumed that the righthand picture is perfectly vertical. The diagram 
concerns 4/7 h — 20 m. 
thesis greater ¢ and « values can be expected than for high 
altitude pictures. In this case even 4 — 0.50¢ is included. 
As is shown the errors for fh = 20 m vary between +2 and 
6 dm (1—3 %). 
It is clear that, even in measuring local height differences, tilt 
may cause errors that should not be ignored. Normally it is not 
clear whether tilt is present 6r not. By measuring vertical pa- 
rallaxes in a number of points in the model it is possible to 
detect and evaluate tilt. Corrections may then be applied to the 
measurements. 
Considerable tilt may be discovered by studying the variation 
of scale in different parts of the picture. The simplest way of 
eliminating the effect of tilt will be to rectify pictures which are 
considerably tilted. 
For the pictures used in this investigation the variation of 
scale in the same picture was highest for pictures in the Ostad 
field, flying altitude 2 500 m. For the pictures information on 
the tilt was acquired by orientation of the model in the auto- 
13