Photogrammetria, XIX, No. 6 
ly a rather limited bx range so that a monocular restitution may not cover the whole 
negative size. 
This does not affect the principle of the use of grid plates but the result will be that 
the unknowns to be solved from the measured coordinates, show greater standard errors 
and less freedom of correlation. This disadvantage can only be partly compensated for by 
including more points in the measurements. However as long as, the practical application 
of the instrument does not include a serious extrapolation outside the range covered by 
the grid measurements, the results obtained will give a fairly good impression of the 
accuracy that can be expected. 
II. 2.2.3. Many instruments are nowadays provided with devices for an automatic 
registration of coordinates on punched tapes or cards. In those cases an elec 
tronic computer will generally be available for the final computations (transformations, 
adjustments a.s.o.). 
Under these circumstances there is no reason to limit the instrument tests to the case 
of vertical projector axis since the computer can easily take care of the more elaborate 
computations due to camera tilts. Different values of tilts can be introduced, the correspon 
ding projected coordinates can be pre-computed and compared with the observed values 
in a way similar to that suggested earlier in this chapter. The advantage of this method 
is that asymmetric positions of the instrument frequently show mechanical, optical, or 
adjustment errors more clearly (especially elastic deformations caused by asymmetric 
loads). 
II. 2.2.4. Stereoscopic grid measurements show several differences with monocular 
measurements, the most important of which are: 
1. The instrument can be used in a position more representative of its practical application. 
2. The introduction of tilts to check to instrument in extreme positions of the projector 
includes less extra computation than in the- case of monocular measurements. In fact, 
the shape of the model should not be affected at all and a similarity transformation in 
space will be the only extra work involved to obtain residual errors in the coordinates. 
3. The influence of residual X and Y parallaxes as differences of the residual projection 
errors in each separate projector are automatically taken into account. 
A disadvantage is, as already mentioned, that during the analysis of the residuals it 
is more difficult (if at all possible) to localize their source. 
As has been recommended for monocular measurements, the stereoscopic measurements 
should preferably be taken in a number of planes. In fact, as illustrated in fig. 1, a grid- 
model has an extension in space which should be used in all cases when the instrument has 
to be used for restitution of hilly or mountainous terrain. The number of points that could 
be used depends obviously on the spacing of the grid lines in relation to the base length. 
For instruments built to execute aerial triangulation, stereoscopic measurements 
should be taken at positive and negative base settings (base inward and outward). 
The combination of these two measurements can give an impression of the systematic 
errors (as far as they are due to the instrument) to be expected in the result of the 
triangulation. Where necessary, a^grid triangulation of more models could be done, with 
the advantage that the influence of random errors in the orientation procedure and in the 
measuring of coordinates can be better separated from the purely systematical errors. 
II. 2. 3. Numerical or graphical output. 
What has been said in the foregoing part of this chapter applies without difficulty 
to those instruments where coordinates can be read directly or where an automatic régis 
tration takes care of this.