• I 
-15- 
Three courses of events are then possible: 
1) Nothing is done about this situation and the results are accepted 
as final. As a consequence, large gaps occur between the models, 
leading to a very low relative accuracy. 
The operators, poor devils, then have to try to match the contours 
and thus a possible gain in relative accuracy is lost. 
2) After the execution of aerial triangulation, consecutive models 
are "corrected" to eliminate these gaps. 
The question is then, why was a method used in the first place 
that would lead to these gaps? Why introduce auxiliary data, if 
their information is eliminated in the next stage? 
3) The gaps and discontinuities are registered and used with proper 
weight in the adjustment phase. This is theoretically correct but 
not very logical. It is certainly easier and safer to execute 
continuous aerial triangulation and to register the discrepancies 
between the instrument data and the corresponding auxiliary data. 
These discrepancies are then used in the adjustment which is the 
best and only logical application of auxiliary data. 
Furthermore, it should be noted that any use of auxiliary data in the 
instrumental phase is certainly time consuming and leads to additional 
sources of errors and that there still remains the complication of 
determining the systematic errors of the auxiliary data. 
The direct application of auxiliary data in the instrumental phase 
might become a necessity if topographic plotters are required for the 
determination of minor control,, In these special circumstances, there 
is no convenient possibility of carrying out strip triangulation with 
the registration of the instrument data corresponding to the auxiliary 
data. This is especially apparent in the case of horizon data. 
This involves, of course, an unavoidable loss of overall accuracy but 
project circumstances can be imagined under which this might be 
acceptable. 
Special care, however, must then be taken to ensure that the results 
of such a procedure are dealt with in a proper way in a subsequent 
adjustment phase. 
One has to avoid any loss of absolute accuracy established by the 
auxiliary data, by applying for instance consecutive relative 
tilt adjustments of models to eliminate gaps (except in cases of 
obvious gross errors).