12. 
From these results it is apparent that the automatic stereoplotting 
device can recover the relative orientation extremely well. At the same time it 
should be noted that the accuracy of the determinations themselves is not very 
good. This is due to the fact that elevations could only be read to an accuracy of 
1l m in the model scale of 1:10, 000 and that only 7 points were used to check eleva- 
tion discrepancies. If more time were available for the test the number of check 
points would have been increased to improve the accuracy of the determinations. 
As a comparison, the same experiments were carried out by an experi- 
enced operator on a Kelsh Plotter in the NRC laboratories. The pointing results 
were recorded by an independent observer and processed in exactly the same way 
as the results of the automatic stereoplotting device. The end results were: 
c 
Ax, = +1.2 
c 
Ad, = 13.4 
C 
Aw, = t1.2 
From this comparison it can be seen that the automatic stereoplotting 
device is superior to the human operator as far as d-values are concerned; only 
slightly superior for x; and somewhat inferior for w. The last-mentioned feature 
may have something to do with the method of introducing the over-correction. * 
Since the vertical parallaxes change little from point to point, vertical 
parallax measurement is a relatively simple task for the device. The automatic 
stereoplotting device is able to gather quickly a large amount of information on 
the vertical parallaxes in a model and it may be expected that its performance 
will be further improved. The information obtained could be best used in an 
instrument that permits a correct statistical treatment of this data. The future 
analytical plotter seems to be the most suitable for this purpose. 
The procedure of orientation in which a bank of mercury switches can 
select the orientation element is very practical. However, it might be possible to 
speed the orientation process by providing the operator with an additional control 
that would allow him to overrule the automatic selection and make his own choice. 
Then it would be possible to eliminate by hand the y-parallax in the initial point 
only and by a proper choice of orientation elements to complete the orientation 
loop with the automatic stereoplotting device in continuous operation. The y- 
parallaxes would then be automatically eliminated throughout the orientation loop 
without a need to make manual adjustments. Since this elimination takes place 
while the scanning head is moved from pointto point a time saving should result. 
Another improvement would be an indicator light to show when the y- 
parallax at a point is completely eliminated. This would save time during the 
orientation process. A still better solution would be an indicator dial to measure 
  
* The inventor of the automatic stereoplotting device has since announced that a 
considerable backlash was discovered in the w-servoing system. This finding 
is supported by the fact that appreciably more time was used for the relative 
orientation towards the end of the experimenting period.