5. Measurements 
The Commission decided to restitute two models out of each flight; each model had to be oriented 
three times independently. After each orientation, the measurements had to be done twice, the second 
time in reverse order. For each model, 5 pass points and 80 check points were prescribed. The report 
on the restituted points indicated whether these check points were situated inside or outside the frame 
of the pass points or in the edge zone of the model (3% of the side of picture). All points were measur- 
ed in advance by a piloting centre (ITC, Delft), which, at the same time, checked the co-ordinates and 
also the remaining data such as point numbers, point descriptions, manner of signalization, etc., 
prior to distributing the information to the restitution centres. Altogether there were 16 co-operating 
restitution centres with the following instruments: 
— Autograph A7 of Wild 
— Zeiss Stereoplanigraph C8 
— Zeiss Precision-Stereocomparator PSK 
— Wild Precision-Stereocomparator STK 1 
— Jena Stereocomparator 1818 
In addition to “general rules for the restitution“, the Commission laid down special “rules for the 
measurements“ [4]. Nevertheless, the restitution centres still had some liberties. It was the intention 
to distribute the tests equally, i. e. the three fold measurements of two models, over the eight flights. 
This could not be accomplished completely. For the analogue machines, for example, the number of 
tests per flight varies between 3 and 8. In the stereocomparators, the models were usually measured 
only once by the restitution centres. Altogether, the measurements of 59 tests, with a total of 290 
models, are available. 
4. Computing Programme 
The original “draft of a computing programme“ presented by the chairman of the Commission 
comprised 10 points. However, only six of these points were incorporated in the final computing 
programme (among other things, the investigations into parallax measurements were left out). If 
the expression “root mean square“ (r.m.s.) error is used here, it applies mostly to “estimated values“ 
in the statistical sense. The computing programme then comprised the following computations: 
(1) Accuracy of measurement (M) 
my is the measuring accuracy of the co-ordinates as derived from the two readings of each point. 
(2) Fitting accuracy (E) 
ma is the r.m.s. error as found from the differences between the terrestrial co-ordinates of the 
five control points and their transformed photogrammetric co-ordinates (after adjustment). 
(3) Absolute accuracy ( A) 
ms is the r.m.s. error computed from the differences between transformed photogrammetric co- 
ordinates and terrestrial co-ordinates of all measured check points (see fig. 1a). 
(4) Relative accuracy (R) 
ma is the r.m.s. error derived from deviations of the three independent measurements of a model 
referred to their arithmetic means (dispersion of co-ordinates as found after repeated absolute 
orientation of the models) (see fig. 1b). 
(5) Accuracy of distances (D) 
ms is also a relative error. Five groups of 20 distances between points were selected, namely short 
and long stretches either in the same model (groups 1 and 2) or in different models (groups 5 
and 4), whilst group 5 consisted of short inclined stretches. In the Reichenbach test area, no 
distances were measured directly; in each case, those required for the comparison were derived 
from terrestrially determined coordinates. 
(6) and (7) are cancelled. 
(8) Systematic errors (S) 
From the differences between the means of the three independent measurements of each model 
(see point (4)) and the corresponding terrestrial co-ordinates, the r.m.s systematic error mg (see 
fig. 1c) was calculated in terms of the quadratic average. 
(9) and (10) are cancelled. 
All restitution centres carried out these computations themselves. 
14