76 RADIAL TRIANGULATION, DISCUSSION 
computed are not errorless themselves. They 
have a certain standard error which is a sort of 
standard error in the directions measured. 
In my report I gave a few examples — six 
examples — of single strips of 48 to 105 kilo- 
metres long in different types of terrain. I 
characterised this terrain by using the well- 
known topographic factor introduced by my 
friend Dr Fagerholm, and these six examples 
you may notice refer to terrain with a topog- 
raphy factor of 0.10 to 1.88. This last type of 
terrain is rather hilly country. There you will 
see that the errors in principal point triangula- 
tion in the first case of practically flat terrain 
are 0.57 metres, and 0.51 metres in x and y 
respectively, for a strip of 48 kilometres. All 
these six strips mentioned in my report are un- 
adjusted. There are no adjustments at all. They 
are single strips, of rather weak construction, 
and I have chosen this construction of single 
strips because it is very easy — or rather easy — 
to compute the propagation of errors as com- 
pared with a bloc, for instance, where the prop- 
agation of errors is much more difficult to study. 
In order to give you another idea, a strip of 
85 kilometres in a country with a topography 
factor of 0.5, a hilly country, gives for principal 
triangulation 7.7 metres and 2.7 metres for x 
and y respectively. In the most hilly country | 
considered, with a topography factor of about 
0.9, these errors were 7.4 metres and 3.5 metres, 
For nadir point triangulation, these errors 
are much smaller. They are, for the three Cases 
mentioned: 0.24 metres and 0.11 for x and y; 
1.27, 1.01; 2.53 and 1.74 metres in x and y 
respectively. It may be interesting to know in 
this latter case, which is the less favourable one 
— hilly country — that this example refers to à 
strip of 85 kilometres; that the errors in x and y 
expressed as a percentage of the length of the 
strip are as follows. For principal point trian- 
gulation in x, 0.064 promille, and for y 0.041 
promille. For nadir point triangulation these 
numbers are 0.030 promille and 0.020 promille, 
For these nadir point triangulations I assumed 
a standard error of the inclination of ten minutes 
which, in view of the recent possibility, is rather 
high, so that these numbers in practice would be 
still smaller. 
Discussion 
Prof W. SCHERMERHORN: One of the im- 
portant problems is that of computation. I draw 
your attention now to a few remarks which are 
TERNURA 
made by Mr Ackermann who has developed a 
new method of strip computation.