The mean square values on the errors on all the control points have been drawn 
from the reports. The data of the last three columns have been derived from table 3. 
The results of these last columns cannot always be compared with those con 
tained in the previous column 7 (m.s.v. discrepancies on all the check points). In 
fact, the number of the points taken into account for the calculation of M m , c m and 
m A does not correspond to the number of the points taken into account by each centre 
for the calculation of the mean square value of the discrepancies on the check points 
of the block. For the choice of the points we followed the method described in 
paragraph 3, chapter VIB), while each Centre followed the different methods described 
in the respective report. The fact that in test 19 the three values of m A are larger 
than the corresponding values contained in column 7 is to be ascribed to the 
above mentioned reason. These last data have been calculated suppressing a 
group of probably anomalous points, as said in the report of the Centre I.T.C. of 
Delft (chapter YE). 
Table 8, even if incomplete and non homogeneous, at least in its first part, is 
very significant. The accuracy we can deduce for the blocks from the data on the 
intrinsic accuracy and from the discrepancies of points used for the adjustment of the 
blocks themselves, is greater than the real absolute accuracy deduced from the coor 
dinates of the control points. 
An exception is made for the mean square values of the differences of points 
common to adjacent strips relative to tests 17 and 18. 
However, these values have been calculated from the coordinates of points com 
mon to adjacent strips after independent adjustment of each strip and before overall 
adjustment of the block. Therefore, those values are not to be referred to the block, 
but to each strip. 
As a consequence of the remarkable differences between the two series of values, 
one can conclude that it is impossible to obtain information on the absolute accuracy 
of a block starting from the data concerning its relative accuracy or, generally, from 
the information one can deduce from the intrinsic comparisons. 
However, before denying any indicative value to these data, we must remember 
that both the errors at the known points used for the adjustment and the discrepan 
cies at the tie points between adjacent strips are the residuals of the equations used 
for adjustment of the block. The mean square value of these residuals can become sig 
nificant if inserted in the more complicated computation of the mean square value 
of the adjusted quantities. 
Therefore, the coordinates of any point of the block will present a mean square 
error which will depend on its position in the block itself, and on the standard error 
of the coefficients of the adjusting formulas. 
Unfortunately, in the present investigation the application of this method of 
evaluation is impossible. In fact, the elements communicated by the centres are not 
sufficient for that purpose. 
Experiments on this problem would be very useful and we call the attention to 
the users of the blocks to this problem. 
180