— Problem 8 
Originally the comparison between analogue instruments and stereocomparators was not included 
in the programme. With om = 2.5, the r.m.s. errors mz in “microns in the image“ for the mea- 
surements with the analogue instruments put together are in the proportion of 1 : V; = 1.00 : 0.65 
to those of the measurements in stereocomparators; consequently the ratio of the height errors 
becomes 1 : V; = 1.00:0.85. The errors m; of the comparator measurements are thus about one 
third smaller than those of the analogue instrument measurements; however, with the height 
errors m;, the differences are slight. 
The r.m.s. errors have often been computed separately for the point groups (interior, exterior and 
edge points), in order to determine the decrease of accuracy towards the edge. Later on, the exterior 
and edge points were taken together, since they were about equally accurate, and the ratios vx and 
v; between their r.m.s. errors and the r.m.s. errors of the interior points were computed. It was 
found that: 
— For planimetric co-ordinates, the decrease of accuracy towards the edge is about v; — 1.55, and 
for heights it is about oz = 1.25. 
— The decrease in accuracy towards the edge is practically independent of the picture scale, of the 
camera used and, above all, of the types of restitution instruments applied (C8, À 7 and stereo- 
comparators) (see [6]). 
In contrast to Oberriet, a hilly, partly mountainous area was chosen for the Reichenbach test, because 
it was suspected that the heights of the points affected accuracy. Finally, the r.m.s. errors of both 
tests were compared and some results of further international tests were incorporated into the com- 
parison, viz: Revoira, Renfrew and Bezons and the Reichenbach test for the ISP. However, it was 
extremely difficult to establish a basis which was to some extent suitable for a comparison of the 
results. Probably this has not always been achieved. With both the Oberriet and Reichenbach tests, 
about equal r.m.s. errors were found. However, if all the other tests are incorporated, then the 
absolute errors m; pertaining to our Reichenbach test and along with these the errors pertaining to 
the Oberriet test, are about 30% smaller than the overall average. For the height error, m;, this 
percentage amounts to 10 only. 
Bibliography 
See French text, page 12.