few checks on transfer from model to model seemed, however, to indicate no loss of 
accuracy. One of the 41 check lengths gave a distance error of + 6,5 cm; a subsequent 
check showed that the control length was wrong! The mean square distance error was 
in fact + 0,6 cm, with a maximum error of 2,2 cm. The taped lengths were thereby 
assumed to be absolutely correct. But this assumption is not justified since the error 
of measurement with the tape is probably of the same order of magnitude. In this 
connection it must be pointed out that photogrammetry gives the same high accuracy 
even for distances of 100 m and more, even in hilly terrain, a precision which is not 
possible with a tape. 
5. Instead of measuring further check lengths in the field, another check was devised. 
The lengths 4, shown in Fig. 7, of the switches, derived from the coordinates, were 
compared with the standard lengths. The lengths of the switches themselves, which 
were too inaccurately defined, were not taken into account. A first check of the 
lengths 4 of the switches showed a mean square distance error of * 1,6 cm with maxi- 
mum errors of + 3,6 cm and - 1,6 cm. A further study of the errors, including a plot 
of the distribution and the probability curve (see Fig. 6) shows that there was a syste- 
matic error of approximately 1,5 cm. Further check measurements in the field of 11 
switches showed that the length 4 was in fact 0,9 cm larger than standard as given in 
the construction drawings. After the standard lengths 4 of 84 switches had been 
corrected in this way a mean square distance error of * 1,2 cm was obtained, with 
maximum errors of + 2,7 cm and - 2,5 cm. 
6. Gross errors 
Photogrammetry is often accused of containing gross errors, though they occur equally 
frequently in conventional surveying. In this project no gross errors could be found 
although approximately 50 % of all measured points were examined in the test. The 
only gross error discovered was the incorrect measurement with the measuring tape! 
Once again it has been shown that a comparison between a photogrammetric survey 
of the highest precision with a conventional survey leads to considerable doubt as to 
which survey checks the other. 
F. Discussion of results 
The results of the tests of accuracy carried out by the Swiss Federal Railways are as 
follows: