148 THE CANADIAN SURVEYOR May, 1960 
5m— 
4m— ] 
3m— 
y 
o 
EI 
Ma T | 7] 7 | lagi] TI 2 
(© Absolute orientation points.) 
Fıc. 5. Elevation errors plotted along a diagonal. 
  
| 
4 
correction and by correcting for the tilt of the stereomodel. 'The results of this 
correction were replotted along another diagonal in the model as a check. 
The results of this operation for all submissions are listed in Table IV. Since 
identical photographs and ground control were used and the elevation readings were 
corrected for all recognizable systematic errors, the figures contained in Table IV can 
be accepted as characteristic for the inner accuracy of the plotting equipment used. 
They are affected of course by the interpretation errors and the personal errors of 
individual operators. These errors are however inherent in photogrammetric operations, 
and when a larger number of results is used the extreme personal errors can be singled 
out and quite reliable average figures can be derived. 
TABLE V—AccURACY OF CONSECUTIVE POINTING FOR 1:50,000 PHOTOGRAPHS ON 
WILD AUTOGRAPH A-7 
  
  
| Operator | 
Time |. °°° °°. S d Mean 
| 1 2 3 | 4 
9:30 | 28 cm 27 cm | 16 cm | 13 cm 20 | 
| | | 
10:30 20 25 | 8 8 15 } 18 cm 
| | 
11:30 | 30 20 | 16 | 8 18 
| | | 
1330 | 17 16 |^ 13 A | 16 ) 
| | 
14:30 | 22 18 | 12 10 15 | 
| | 16 cm 
15:30 | 7 1 | ^12 10 15 
16:30 | 20 | 38$ 15 8 16 
. — . : 3b : | 
Mean | 23 cm | 21 cm | 13 cm | 10 cm 
It is noted that the results in each instrument group are quite consistent and show 
a definite improvement over the results in Table II. A detailed analysis of each 
submission proved that the differences between the results in Tables II and IV are 
mostly caused by human errors which may on occasion amount to appreciable size. 
POINTING ACCURACY AND HUMAN Errors 
Pointing accuracy is one of the basic factors which directly affect instrumental 
readings. ‘The repeatability of pointing in elevation can be readily established by 
reading a point several times and by computing from the differences, for instance, the 
mean square error of a single reading. Care must be taken to keep the operator 
ignorant of the reading values, otherwise the readings may become more consistent 
than they should be. In our experiment four persons participated in the determination