dicular to that of the line feature itself, may be of the order of 2 to 50 
cms, depending on the nature of the boundaries. Idealisation errors of 
properly signalised points, however, are practically zero. 
1.4.1.1 Numerical Restitution of Signalised Points 
On the accuracy of numerical stereo restitution of signalised points 
there exists a wealth of experimental and production data. From the results 
of Commission C of the OEEPE (European Organisation for Experimental 
Photogrammetric Research) of the Reichenbach Experiment (ref. OEEPE 
Pub. 4, December 1968), for instance, the following accuracy data emerge. 
Using five (signalised) ground control points per stereo model for 
linear conformal transformation of the model co-ordinates of a large 
number of signalised check points to the ground co-ordinate system, the 
following standard deviations were found in the planimetric co-ordinates 
(0, =0 y^? c of, and the distances ( 0 à) between, the signalised 
points: 
0 = 12.5 microns (at negative scale) 
0 q = 11 microns 
short 
0 q = 15 microns 
long 
This was for wide angle photography (9” x 9”, 6") at scales of 
1:8,000 and 1:12,000, for numerical restitution on analogue instruments. 
For numerical restitution on a Stereo-Comparator, a 20—25% better result 
was obtained. These results are based on work carried out by the many 
European organisations participating in the experiment, using different 
makes of instrument. 
It goes without saying that for cadastral purposes and large scale 
restitution in general, the relative accuracy, i.e. the accuracy of distances, 
is much more important than the absolute accuracy, i.e. that of co-ordinates. 
This is acknowledged by existing accuracy specifications which call for a 
certain accuracy of distances. The short distances in the Reichenbach Expe- 
riment are distances of between 40 and 240 metres (up to about a quarter 
of the airbase B), the long distances are from 250 to 575 metres (roughly