This capability is also an peculiar characteristics of analytical method
which manifest itself in a flexibility of the method. Therefore, there
are important problems relating to remeasurement, 1) how to detect wrong
points to be remeasured, and 2) accuracies of remeasuring the 4 fiducial
marks at the corner of a diapositive which are the reference points of
remeasurements.
In the last chapter, we have described the general feature of our
system. It is needless to repeat that our system is structured aiming
at mainly to be convenient for the detection of wrong points, that is
for the convenience of remeasurement. Therefore, our computing system
is the one which concentrates its purpose to the former of the above
problems. About the latter we will mention here in some details.
We have not made special experiments to test the accuracy of re
measurements of the fiducial marks. But from the data which are being
produced automatically day by day as the necssary results of program 1,
we are able to make a statistical look on this accuracy.
Now we have at hand over ten thousands of examples for measured
values of fiducial marks, thereby the camera and focal length are of
various kinds. And 20 or J>0% of them are being remeasured. Since
complete statistics is much laborious and time consuming, we have made
a statistical treatments for about 20% of the total as below.
First we will show some characteristics of the content of this
statistics in Table 1. Materials are classified according to 1) the kind
of c a mera, that is, the camera No.1 has 4 fiducial marks at each middle
point of lateral sides with the form of sharp wedge, and camera No.2
has those of sharp crosses at each corner of photograph, and 2) the kind
of measuring instruments, A7 and comparator. We see in this table that
the occurences of remeasurements are remarkably lower in the case when
the measuring instrument is A7. This will be almost due to the better
quality of optical system of A7 than that of our comparator, but partly
due to the fact that tie raw materials by comparator may be brought from
those which were made by not skilled operators- at least it is true
that our comparator has yet short history. In table 1 it is shown that
in total about 20% of photgraphs are remeasured and 10% are remeasured
more than two times. Now we are thinking about that we may have a hope
of reducing the percentages of remeasurements to be lower than 10%.
Now we have the other statistics. This statistics was made directly
at the section of measurement. During the period from April 1963 to
8