1/4 — 2/3 B). Had there been no correlation between the errors in the two 
points which form a distance, then the 0 d Would have been equal to 
v/ 2.0 ,, thus 17.5 microns. 
For extremely short distances (where d 7 0), 0 q Will only include 
the effect of the setting errors of the two points, because in that case the 
instrumental errors and photo errors in both points are practically identical, 
and thus od = /2.(0 e setting: In the Reichenbach Experiment it was 
found — from repeated settings — that ( 0 e = 4 microns, and hence 
setting 
0 d £ 6 microns. 
We therefore have, for the Reichenbach experiment, the following 
relation between errors in distances and the distances themselves. 
Shorter distances have significantly smaller errors than longer distances, 
thanks to the high positive correlation in the stereo model. This is, however, 
only true for distances with end points in the same stereo model. In the 
Reichenbach experiment it was found that short, as well as long distances, 
with end points in different stereo models, had a standard deviation of the 
order of 17.5 microns. This is as large as one would expect for errors in 
very long distances with end points in the same model. Obviously this is 
caused by the effect of different instrumental errors and different photo 
errors in the two end points situated in different stereo models, and by the 
unavoidable closing errors in fitting each model to its ground control points. 
  
  
  
Vo 
T | 
%B NB 
figure 1