E.H. Thompson. There are enough check points to make the difference insignificant. 
E.H. Mikhail (on behalf of D.C. Brown). Percentage drop in standard deviation is unrealistic, the percentage 
drop in variance is more realistic. Also, iteration may not have terminated, in some adjustments. 
G.H. Schut. As participant 1, I did not consider it realistic to adjust without the two extra controls in the edge. 
I consider the difference between participants 8 and 10 (Tables 2B and 2C) significant and worthy of further 
investigation. The comparison of polynomial adjustments requires care because of variation of methods. The 
use of secant plane heights rather than terrain (geoid heights) tends to give a favourable bias, since their 
computation is not always practicable. 
H. Ebner. Methods of block adjustment should be considered in three categories: strips (polynomial), models 
(simultaneous absolute orientation of all models in a block) and bundles of rays (simultaneous adjustment 
photographs). With comparator observation, bundles ought to be used, but the advantage of the model method 
is that either comparator or analogue plotter observations may be used. The Stuttgart model adjustment 
PAT-M, costs less that $1/model. 
J.B. Case. The theoretical advantage of the simultaneous (bundle) method is that it can make better use of 
realistic weights, auxiliary data and allowance for errors of secant plane conversion. 
F. Ackermann. The interpretation of the results presented should indicate whether they agree with theoretical 
expectations. The comparison of analytical and model methods agrees with our theoretical findings, and 
generalizations may be made concerning the random error case. However, I raise a warning when systematic 
errors are present, where the comparison may vary. For example, in the OEEPE test block of about 200 actual 
photographs (presumably with some systematic error), bundles gave a worse result than independent models, 
contrary to theoretical expectations. 
Friday July 28 1972’ 9:00 
Invited Paper: “ISP Investigation into the Accuracy of Photogrammetric Triangulation"' by K. Kubik and J. 
Kure. 
Panel discussion 
D.W. Proctor. I ask the panel to consider, not only the results tabulated in the report, but the whole question 
of further research into block adjustment and the computations which precede it. We have had reports now 
on tests with simulated data, and of practical adjustments, and must consider the future programme of 
Commission III in this direction. 
F. Ackermann. Considerable work has been done on research into theoretical accuracy studies and controlled 
tests (i.e. special or simulated photography). This should immediately lead to a new programme of work for 
the next congress. Included in this should be statistical tests. 
Analysis of practical results might pause for a year or two, as methods are currently in a state of change. 
At Stuttgart, about 30 blocks have been adjusted for cadastral purposes with consistent results: independent 
models observed on analogue instruments gave 0, in the range 6 to 12 jum, stereocomparator observations 5 
to 7 pm. These are all signalized points. With artificial marks for pass points etc., 0, becomes 15 to 25 pm. 
The significant difference is the use of signalized points. A block adjustment process is used, even when there 
is enough control for each model to have an absolute orientation. 
G. Togliatti. The data in this report is only significant in representing the present state of affairs. The analysis 
inevitably considered non-homogeneous data, and extrapolation of the results would be dangerous. I agree that 
theoretical studies should continue, and any further practical analysis should be very selective. 
J. Kure. A number of responses were excluded, owing to unreliability of the data. One factor is that some 
organizations do not need the highest obtainable accuracy, and are therefore justified in using less refined 
techniques. Some use very short bridging distances, 2 or 3 models, in some cases supplying 16 extra controls 
for the purpose; this is hardly aerial triangulation as we would understand it. 
J.M. Anderson. Theoretical and practical studies should both continue. If the start of the analysis is delayed, 
I know from experience that there will be difficulty in obtaining results before the next Congress.