32 
of random noise. This fact--leaving aside economical con- 
siderations--may prove to constitute a major challenge to 
the design and manufacture of completely automatized measur 
ing equipment. 
The 930 single camera reductions--ranging from about bo 
to 100 plates per lens cone—demonstrate, with the quanti 
tative terms of their specific scale, radial and decentering 
distortion parameters, that the photogrammetric bundle can be 
simulated reliably by a specific mathematical model over a 
temperature range from +20°C to -36°C. The similarity of 
these parameters, when normalized for temperature influences, 
gives evidence that the series of 1? Wild BC-b Cosmotar cones 
were manufactured to extremely close tolerances, resembling each 
other closely and performing close to theoretical expectations. 
The application of the analytical reduction method to the 
calibration of wide-angle precision cameras has so far pro 
duced residual errors compatible with the results obtained 
during the data reduction of the satellite triangulation pro 
cedure. This fact suggests that the theoretical potential 
of aerial triangulation can be realized, provided proper 
precautions are taken to control or monitor the environ 
mental conditions of the aerial camera during flight. 
The contribution of random errors of the 103E emulsion 
on glass plates was statistically isolated and determined 
to be +1.0 micron, a result which agrees exactly with the 
result of independent laboratory tests made by Eastman-Kodak. 
Atmospheric scintillation, with an average of ±2.1 micron 
(±1"0) for star and ±2.7 microns (±1"2) for satellite images, 
constitutes one of the major error sources in photogrammetric 
satellite triangulation. The performance of a comparator 
with respect to its random behavior therefore need not be 
better than ±1.0 micron for the type of application under 
consideration. 
Following re-evaluation of all presently available and 
future raw data with a new generation of existing computer 
programs, more statistical analysis, involving correlation 
techniques, will be applied, designed to investigate the 
components of the photogrammetric procedure in finer detail. 
As a by-product of the program information will become available 
in areas outside of photogrammetry. Examples are: the quan 
titative determination of electromagnetic propagation delays 
by comparing received time signals with the station clocks 
whose performances are monitored by frequent comparisons,