two instruments (Zeiss Ortho-3 and Ortho SFOM 9300), the orthophotos 
originally submitted were positive transparencies and they were used 
directly in the accuracy evaluation. As noted in Table I, the scales 
of orthophotos varied from 1:2 500 to 1:10 000. 
Diapositives produced at a scale of 1:2 500 on dimensionally stable 
film from those orthophotos which were not originally submitted at 
the scale requested. A Borrowdale cartographic camera was used for 
the photographic reduction. 
Transparent reductions at a scale of 1:10 000 of all the orthophotos 
submitted, with the exception of the Zeiss GZ-1 orthophotos which 
were reduced to 1:7 500 in view of constant reduction factors used 
in the project. The Borrowdale cartographic camera was used to 
produce the reductions. 
The full size transparencies at 1:2 500 scale of each of the 
orthophotos were measured on the NRC Stereocompiler II, an instrument 
developed for plotting and measuring of stereo-orthophotos. This instrument 
is equipped with Tinear encoders to measure the X, Y orthophoto coordinates 
and X-parallaxes. 
In addition to determining the accuracy of the individual 
orthophotos by measuring the coordinates of control points, the Stereo- 
compiler offers the possibility of viewing the orthophoto pairs stereo- 
scopically and measuring the X-parallaxes. Such a stereoscopic inspection 
immediately discloses the inaccuracies in the differential rectification 
process without the need for control points and coordinate measurements. 
Ideally the stereomodel composed of orthophotos, made from two consecutive 
photographs, should be flat, with the exception of "vertical features" 
such as trees and buildings. Any geometric imperfections in the 
orthophotos, however, will produce bulges or depressions in the stereomodel 
which are easily spotted and measured in terms of variable horizontal and 
vertical parallaxes. 
The accuracy of the Stereocompiler measurements, tested on a 
suitable 350 mm x 700 mmgrid, proved to be +18 um for the X, Y coordinates 
in the scale of the original photographs and +10 um in determination of 
horizontal parallaxes. This was thought to be somewhat restrictive. For 
this reason all orthophotos were reduced to the scale of original photographs 
1:10 000 using a cartographic camera. This permitted the test measurements 
to be carried out on the NRC Monocomparator offering a measuring accuracy 
of 1.5 um. However, in the latter case some additional errors were 
introduced by the reduction process (film deformation, projective errors, 
etc.). By reducing the above mentioned grid in the Borrowdale cartographic 
camera and measuring the reduced grid on the Monocomparator, it was found 
that this evaluation process can offer an accuracy of my = 10 um and My 
= 8 um, which is better than the accuracy provided by Stereocompiler 
measurements. Consequently, the Monocomparator measurements of orthophotos 
reduced to 1:10 000 scale have been used as the primary data in establishing