The statistical checking succeeds in two subsequent steps. 
At first it is investigated, whether the corresponding correc- 
tion terms of the different parameter groups differ significant- 
ly from each other, If this is not the case the concerned para- 
meters are combined to one parameter, Thereby the individual 
correction terms (bj, bo, ...) can be treated separately. This 
is possible because the additional parameters are nearly ortho- 
gonal and influence each other only slightly. With the corre- 
spondingly combined parameters the block adjustment then is re- 
peated. 
In the second step the significance of the remained correction 
terms is checked and weight e (10?9) is given to the insigni - 
ficant parameters, A last repetition run then gives the final 
results, 
It easily can be realized, that both steps of this procedure can 
be performed automatically by the program, so that the program 
user has not to be burdened by the valuation of the obtained 
Correction terms and the according consequences, 
The sophisticated two step procedure, suggested here, guarantees, 
that only the well determined additional parameters are finally 
used. In that way optimum reliability is attained. 
Of course, the computing time of block adjustment increases when 
simultaneous self calibration is implemented. In practice how- 
ever, the additional amount will not be very high, because the 
correction terms usually will only be put up with the very last 
runs, whilst the first runs by which the gross errors of the 
data are detected and eliminated, will be performed as up to now, 
A. TEST. RESULTS 
For a practical test of the suggested strategy a part of the 
test block Oberschwaben was used. The author appreciates that 
the OEEPE has made this valuable material available. The chosen 
Sub block is built by the wide angle strips 5, 7, 9 and 11 and 
consists of 100 models (terrain area = 20 . 62.5 km2). All con- 
trol points and tie points were targeted. The flight was per- 
formed with a Zeiss RMK A 15/23 camera at a photo scale of 
1:28 000. A Zeiss PSK stereocomparator was used for image 
measurement. 
Starting from the same image coordinates the block adjustments 
were performed by bundles and by independent models. In the 
second case the models were formed computationally. 
The bundle block adjustments were computed at the Technical 
University of Munich, using the self calibrating program, deve- 
loped by Dr. Grün |15|. The flexibility of this program with 
respect to the functional model allowed to use exactly the 12 
additional image parameters, suggested in 3,1, The block adjust- 
ments by independent models were performed at Stuttgart Uni- 
versity, using the self calibration program of Mr. Schneider, 
This program was developed to test the concept of simultaneous 
self calibration before an integration into the program package 
PAT-M |14|. For the given support the author is grateful to 
Dr. Grün and to Mr. Schneider.