72 RADIAL TRIANGULATION, ROELOFS 
no longer a problem, the I.T.C.-Jerie method of block adjustment being applicable. This 
is of extreme importance, a block adjustment not only serving the adjustment of acci- 
dental errors, but also being capable of reducing considerably the effect of systematie 
errors such as given in the table for a single strip. As a matter of fact a large block of 
R.T. is being adjusted now by this method at the Geodetic Institute of the Delft Tech- 
nical University. It is hoped that information on the result obtained will be available at 
the London-congress. 
I A peculiar method of R.T., the *T-triangulation", designed by Sutor [4], was sub- 
Im d mitted to a practical test [5]. The T-triangulation differs from the classie R.T. in that 
Im the sides of the polygon of radial-points are determined, not by angular measurements, 
but by measuring in each photograph the lengths of two lines in a *T"-shape: a side of 
the polygon and a line perpendicular to it in its terminal point. Practical application by 
Rauhut to two adjacent strips of nine photographs and 3.5 km length, gave a standard 
deviation in position of 1.55 m, computed from the discrepancies of 28 check-points. 
Il A method which has some similarity with the preceding one, is called Photopolygono- 
MN metry and was designed by Skiridow [6]. Here the sides of the polygon of nadirpoints 
are measured directly in the photographs and reduced for camera-tilt. Application by 
Rauhut [5] to the same example as in the preceding case gave a standard deviation in 
position of 1.45 m. 
Another variant to the classic R.T., designed by Hallert [7] and [8], consists in the 
measurement, in each pair of photographs, of rectangular coordinates — instead of direc- 
tions - with respect to a system of axes through the principal points and along the base 
| lines. Ground coordinates are computed by intersection from the principal points. From 
| identical points in the overlap of successive photographs, scale and azimuth are transfer- 
hill red by coordinate transformation. An obvious disadvantage of the method is that it 
I requires the measurement in each photograph of 18 rectangular coordinates or parallaxes, 
Ln whereas in the classic R.T. the same result is obtained by measuring only 8 directions. 
The fact that the computations directly refer to rectangular coordinates has ceased to be 
an advantage since Ackermann’s investigation [3] has given the same possibility for 
H | M classic R.T. The method was applied to a simple test case of only two photographs, 
| photo-scale 1 : 43500, to give a standard error in X and Y of about 1 meter, computed 
from the discrepancies at seven check-points [9]. 
  
  
  
  
  
  
  
3. Mechanical Radial Triangulation. 
This method, with which far more and larger areas have been bridged than with any 
other method, either radial or spatial, is so well-established in practice and has been 
submitted so often to extensive and deep-searching investigations, that many new devel- 
opments could not be expected. Mention must be made however of a method, disigned by 
  
I Visser [10], to use local conglomerations of points, whose scale and azimuth are deter- 
| mined by terrestrial-astronomical methods, as control points for the slotted templet 
ili triangulation. For each local system of points a block templet is made on the basis of a 
separate assembly of the templet covering the area in question. These block-templets 
Mb substitute the templets of the local areas in the lay-out of the whole area to be triangu- 
lated. A simple parallel-guidance allows the block templet to shift in both X and Y direc- 
il tions, while maintaining its proper azimuth. Since the position of the local systems of 
| m points need not be determined, but just scale and azimuth, the method seems to be very 
BT 0 1 practical. 
ill | il A more or less new development, by Konecny [11], is the R.T. applied to convergent 
i it photography. Three methods are proposed: isocenter R.T. with contact-prints, nadirpoint 
il i R.T. with approximately rectified prints (both methods either with slotted templets or 
  
  
  
  
| metal arms) and the stereo-templet method. From a practical test the following con- 
ii clusions are drawn: as to the precision the metal arm system falls far behind the templet 
  
  
  
  
  
  
  
  
ees 
— — MÀ