of 20, 000ft. will make the centre of the overlap 7ft. higher than the extreme 
corners and 4ft. higher than the corners of the normal working area. The ten- 
dency of the overlaps to set too high when including the results of the high-level 
profile may thus be accounted for by curvature effect. 
With low-level APR each of these causes of error is greatly minimised. There 
will be moreover far more precise identification of the 35mm p. p.s on the survey 
photographs. 
For these reasons it was decided to discard the results of the high-level 
APR altogether and to use only the points selected from the profiles running 
along the lateral overlaps of the survey photography. This still left at least 
six points (three from each independent profile) on which to level any overlap. 
As an additional check there was the carry-forward from one overlap to the next. 
Fifteen clear-cut cases of side-lobe effect were discovered also in the low- 
level APR, allin areas close to the river bank or in the margin between dry and 
damp ground. Another seven points which showed residual levelling errors 
between 10 - 40ft, proved on examination to fall on abrupt slopes or in clumps 
of dense trees and should therefore have been excluded during the earlier analysis 
of the 35mm p.p.s.. It is however important to note that all these twenty-two 
false values showed up 'like sore thumbs' at the stage of setting the overlaps. 
Of the remaining 1, 040 points used, the maximum residual error in the over- 
lap was "ft. and the r.m.s. error 2.1ft. These are not of course absolute 
errors, but are an indication of the frequency and magnitude of random errors, 
besides demonstrating the high degree of sympathy held between groups of 
points taken from independent APR profiles. 
Many of the overlaps contained points which had been identified on the 
photographs and spirit-levelled on the ground during the large-scale mapping 
at the dam site and of Yelwa Town. In addition there was continuous ground con- 
trol all along the road strips which ran both sides of the river and often followed 
close to the limiting contour of the reservoir. Excluding the points which fell in 
the control zones, none of these additional levels were used in the initial setting 
of the overlaps but were held in reserve as an absolute check on the final 
accuracy of the work. In one or two cases very minor adjustments were made 
to improve the readings at these additional ground control points before plotting 
the contours but the errors listed in Appendix II and shown in Diagram B are 
those occurring when comparison was first made with the check levels. They 
may therefore be accepted as the absolute errors resulting from the use of APR 
with no other control than that provided at the points indicated in Diagram B. 
11