388 
plotting. Unfortunately, such an idea, while simple, has practical limitations. The 
area needs to be large, with closely spaced and clearly marked ground targets, and it 
is therefore expensive to set up and maintain. In view of the many variables— 
temperature, meteorological situation, film flatness and processing, aircraft environ 
ment, etc. it is not certain that corrections obtained from one flight could be used 
in subsequent flights over widely different terrain. 
Nevertheless, such a test area can give valuable information. For research 
purposes, repeated flying and proper statistical analysis could help to isolate the 
various sources of errors and allow some badly needed quantitative data on them to 
be obtained. Such an area could also help to “calibrate” the aircraft installation, 
when new aircraft are commissioned. A few flights over the test area would soon 
indicate whether there were any serious faults in the heating, windows, turbulence, 
and so on, and save a great deal of time and money on wasted photography or 
difficult plotting. Hallert [22> 91 in his communication to the 1960 Congress and a 
subsequent paper has given some detailed results from such a test area and also from 
a high television tower, where the results are free from some of the aircraft problems 
such as movement and vibration. 
18. Work to be Done 
In general terms, I believe that the first essential step to progress in reducing the 
errors in the photograms is to obtain more quantitative information. This informa 
tion to be useful must come from properly controlled experiments, the measurements 
must be subject to statistical evaluation, and the reporting must define the experi 
mental conditions, the terminology used and the accuracy of the measuring equip 
ment. Much of the literature leaves very much to be desired in all these respects. 
In more specific terms, I would suggest that we need: 
(a) More quantitative information on the accuracy actually obtained in survey 
photography, deduced of course from subsequent plotting. Such informa 
tion should give not just a single overall figure for the error, but also the 
distribution within the format and the direction of error at each position. 
Additionally, all the experimental conditions need to be stated, the camera 
calibration, the film, the processing history, the method of correcting for 
distortion, film shrinkage, etc. A systematic analysis of a number of 
negatives might yield valuable information. 
(b) An increase in the number of test areas available to air survey organisations, 
so that the whole aircraft installation as well as the camera can be calibrated. 
These test areas would also be invaluable for research into sources of error. 
(c) More detailed calibration information from a wide selection of cameras, 
with more than the four diagonals so that distribution of distortion 
throughout the entire format can be deduced. All the experimental 
conditions should be stated (including accuracy of measuring equipment) 
and the results either given in their original form as measurements or after 
systematic treatment by the method of least squares. 
(d) An investigation of tangential distortion. First, the designers of lenses 
could justify the “equivalent prism” concept by ray tracing for modern 
lenses. Secondly, we should have more practical measurements of tangential 
distortion, so that these can be compared with the theoretical values 
obtained from the “equivalent prism" theory and the asymmetry deduced 
from separation of centre of symmetry and principal point of auto- 
collimation.