3
accuracy. It is not, perhaps, irrelevant that Boniface’s results
(with and without the reseau) were extremely good (Eden I967) and
that therefore the improvement found by Sitek may have been due to
correction of geometrical distortions in the particular set of dia-
positives used by him. If so, it is none the less a credit to the
use of the reseau which is thus seen to be a permanent control on the
distortions of relatively old negatives.
Further tests were carried out by T.J. Blachut (1966) with a
different camera (and different reseau system) which showed very great
improvements when the reseau was used. This was ultimately ascribed
to correction of the distortion due to serious lack of flatness of
the film; and the possibility of doing this is one of the advantages
claimed for the reseau - but, of course, only when it is projected
centrally and not orthogonally on the film.
What is clear, however, from all these tests is that
asymmetrical lens distortion can cause greater errors than any arising
from the film. Fortunately asymmetrical distortions are easily
corrected when the plates are directly measured (by mono- or stereo
comparator). I may, however, be unfair when I add that there is a
noticeable tendency on the part of manufacturers to suppress the
evidence of asymmetrical distortions, by publishing distortion curves
that are the average of measures made across several diameters* I
see no reason for manufacturers to be ashamed of asymmetrical distor
tions several times an average figure. The modern objective gives
remarkable resolution at a large aperture and angular field. But, as
a direct consequence of the complicated designs that are necessary to
achieve these qualities, it is extremely difficult to make. By
suppressing the asymmetries that arise during construction the manu
facturer is making a stick to beat himself with because, as
analytical methods become more widespread, more and more photogram-
metrists will find out the defects for themselves.
I now pass to a problem that not only affects accuracy but
also efficiency and cost. The problem of whether it is best to
observe in a mono- or a stereo-comparator. * I have personally tended
to favour the latter but now I am not so sure. Let me set out the
advantages of stereo-observation. First, for relative orientation
measures, no marking of the pictures is necessary at all and conse
quently any errors that arise from marking do not affect relative
orientation upon which the accuracy of aerial triangulation basically
depends. Second, for other measures (joins to adjacent models and
strips), it is necessary to mark one picture only and although the
one mark may tend to spoil the stereoscopic vision it will not do so
if carefully made and is the right size. What, however, is
important is that its location is not critical. Neither of these
advantages seemed to be shared by measurement of single plates, but
some recent work by Eden described in the paper already quoted has,
* or, rather, in a mono- or stereo-manner for there is nothing in a
stereocomparator that prevents it being used as a mono-comparator.
+ however the matter is disguised in the computation.