Iso 
rn 
ap 
ric 
Ors 
on 
Iso 
Zà- 
gle 
ral 
els. 
ing 
ere 
ta- 
era 
ork 
ith 
ted 
vey 
out 
nce 
the astrofixes have unknown deviations of the plumbline. Three new bases were established 
whose length and azimuth were determined with a relative mean square error in x and y 
of about 0,5 meters. Finally a few control points were derived from a previously executed 
slotted templet lay-out whose relative mean square errors in x and y are estimated at 
about 25 m. It is obvious that for this river survey the ground control has not a homoge- 
neous character. 
TABLE 4 gives in an excellent way a description of the triangulation of a number of 
blocks executed by this organization. No further explanation is necessary. We draw atten- 
tion to block I a and I b mentioned before, which proves that with rather poor ground 
control a reasonable result can be obtained if sufficient camera orientation equipment is 
used. In combination with block adjustment it was, in this case in particular, the airborne 
profil recorder which was a great help, although during the first season trouble with this 
instrument resulted in quite a number of filling strips without A.P.R. registration. Since 
this aerial survey did not require greater precision than for a map 1 : 100.000 with contour 
line interval of 50 metres, the block adjustment could be carried out with sections consisting 
of more than 2 models; 3 strips of 6 models each were combined to one section. This 
reduction of the scale of the sections has naturally an unfavourable influence on the 
residual errors in the tie-points, expressed in microns in the scale of the negatives. If in a 
normal case the mean square value of these residual errors would be 20 microns, we can 
expect with these reduced sections between 50 and 70 microns. In addition to this comes 
the inner deformation of each of the sections caused by the composition of each section 
out of 18 models. This deformation which cannot be avoided will remain after the adjust- 
ment and influence unfavourably the residual errors. All this could be accepted, however, 
because of the small scale of the map required and the corresponding large tolerances. 
That the requirement in this case could be fulfilled in a very satisfactory way is proved by 
the relative mean square error of 5,8 and 7,8 meter for the two blocks in x as well as in y, 
which means in the map 0,06 and 0,08 millimeters, values which are beyond the possibil- 
ities of a draughtsman. The absolute mean square errors are in this case not of great value 
since these are derived only from the residual errors in the control points. Since there are 
no other known points available in the entire area, a mean square value of 15.4 and 16,0 
meters for a difference in coordinates with astro stations proves only that there are no 
large deviations of the plumbline in that area. It has to be mentioned that in these blocks 
the ground control was of a poor quality. In block Ia 4 astro fixes with surrounding base 
lines were used, along the north boundary of the area. These 4 points gave a mean square 
value of the residual errors of only 12,2 m. Along the southern boundary with Brazil they 
tried to identify points of the watershed-boundary determined by a boundary expedition 
about 25 years ago. This succeeded quite well although not everywhere with the same 
certainty. Also the precision of the boundary survey is not known sufficiently. One point 
misclosed by several hundred meters and was rejected. The remaining 8 points gave 
together a mean square value of their discrepancies of 76 m. Of these 8 points 1 was used 
as an isolated point and the other 7 were combined into 2 groups of points of which the 
centres of gravity were introduced as known points in the adjustment of block Ia. This 
made together with the 4 northern astro fixes the 7 ground control points which gave mean 
square values of the residual errors of 15,4 m in x and y. For the evaluation of the usefulness 
of the analogue computer for block adjustment, this case of the adjustment of about 
2750 models in one block and the resulting mean square value of the residual errors in 
the tie-points is of much more importance and convincing about the great value of this 
analogue computer for the solution of such a problem. Considering however the fact that 
as ground control for the eastern block Ib 3 points were used from a french survey along 
17