The second publication, while primarily concerned with a description of the
procedure of observation and subsequent calculations, gave the results of two
blocks: one using the reseau, and the other without it. Since no reseau block had
been completed at the time of its writing, it gave the results of the block of
photographs without the reseau and „therefore containing all the non-systematic
errors due to film distortion and lack of flatness which the reseau method
eliminates"; the results of a block consisting of photographs carrying the reseau
was given at the Meeting at which the paper was read. The results of the two
blocks are reproduced in Table 11.
Table 11
non-reseau block | reseau block
Number of photographs 210 | 229
Area 240 km | 145 km
Number of Trig. stations |
used in the adjustment | 10 | 18
Number of photographs 21 | 12 (Approx.)
controlled by one Trig. |
station |
! Number of check points 10 | 29
| Standard error at check Northing .85m
[s vector .65 m
points | Fasting 71m
|
It seemed to me however that both the wording of the paper and the present-
ation of the results as above might give the impression that the improvement had
entirely resulted from the use of reseau "because it eliminated the distortions and
lack of flatness of the film.‘ The critical reader will not however fail to notice an
excess of ground control in favour of the reseau block. But I thought it was not
possible to learn from the text that twelve minor control points were observed
and used per overlap in the reseau block whereas the non-reseau block had been
observed and computed according to the old 6-point method. Other factors in
favour of the reseau block were also not clearly specified: There were the differ-
ences between the methods of calculating the relative tilts to which I referred
earlier; and corresponding differences in the methods of joining successive stereos-
copic models; as well as improvements in the methods of adjustment. It is because
these differences between the methods of treating reseau and non-reseau photo-
graphs had not been clearly specified that the paper was likely to mislead.
6.4. The next question is how much more effort was needed to sort out the
contributions of these factors, thereby making the work of greater value.
To fix ideas let us suppose that the methods of computing and adjustment
had been standardized so that one could safely assume that the subjects of the
investigation were only (a) the use of reseau versus collimating marks, and (b)
the use of double versus single minor control.
Now, to suggest to reobserve and recompute the non-reseau block would be
practically unacceptable, since it would involve the unwelcome repetition of work
already passed and included in the production. This repetition is also unnecessary
24