The differences between the A-7 and the automatic contour lines are 
amazingly close to the theoretical error values, in spite of the systematic shift 
between the two plots. The planimetric error of contour lines decreases with the 
slope angle of the terrain, although slower than for a human operator. 
The automatic contours are generalized and in this characteristic they 
are similar to contours produced in second- and third-order plotters. Some of 
the contours are missing, but this is the fault of the operator who did not perform 
the final check for the completeness of automatic contouring. 
In addition to the foregoing analysis of the accuracy of contour lines 
8-km long vertical profiles were constructed from both plots at 1:10,000 along 
identical straight lines (Fig. 15). 
The systematic shift between the two plots is clearly visible. The 
largest discrepancies occur within forested areas, where human contouring is 
also not too reliable. The average discrepancy for the entire profile between the 
automatic and the A-7 profile amounts to t3. 8 m, which coincides with the value 
that could be derived from data in Table III. 
Repeatability 
In order to gain some information on the repeatability of the automatic 
contouring the 150-m contour was plotted three times at twice the speed of human 
plotting (see Fig. 16). From the results it can be concluded that the repeatability 
is very high. With the exception of very flat areas where planimetric displace- 
ments do not produce appreciable errors in elevation and which are also difficult 
to contour for a human operator, the occasional planimetric discrepancies do not 
exceed3 mm at a scale of 1:10, 000. The corresponding maximum elevation 
differences are smaller than 2.5 m, which is an excellent result for the type of 
terrain involved and the 1:50, 000 scale of the photographs. Mostly, however, there 
is a perfect agreement between all three plots. For the purpose of comparison 
the same contour line was plotted three times by the same operator on a Kelsh 
plotter. In spite of the fact that the operator was using his memory in each con- 
secutive plotting operation as to the location of the contour line, his repeatability 
in contouring was very much the same as that of the automatic device. 
Contouring Speed 
  
So far the contouring test was conducted at a speed which was twice the 
peak contouring speed of a human operator. The peak speed of the operator was 
determined from the uninterrupted plotting of a single contour line. When the time 
necessary for contouring larger areas is compared, the speed of automatic con- 
touring will be about three to four times the speed of a humanoperator. It is quite 
obvious that the human operator cannot maintain his peak speed over a long period 
of time. The actual time used for the contouring plot, as shown in Fig. 11, was 
two hours and five minutes. The shortest time used on identical contouring by 
one of the 9 Kelsh plotter participants in the international plotting experiment was 
17 hours and the average contouring time of the 9 plots was 32 hours. However, 
in the international experiment twice as many contours were plotted, and so the 
time valid for comparison is 8. 5 and 16 hours, respectively, In addition, about 
one to two hours should be added for the checking and possible correction of the