SOME EXPERIMENTS OF SEMI ANALYTICAL TRIANGULATION 
13 
from some tests we made, we can deduce that a skilful operator could perform the 
computations in the same time the models are formed and surveyed. 
In this connection, we must point out that semianalytical triangulation re 
quires a greater number of computations than that necessary for analogical trian 
gulation. This is to be taken into account when the method is used with desk ma 
chine computations. 
Viceversa, if the computations are executed in an electronic way, the time 
required for the connection is quite négligeable (also from an economic point of 
view), especially if we consider that, at any rate, the adjustment computation 
either of the strips or of the blocks, is always necessary. 
In connection with the use of an electronic computer, we must outline the 
other possibilities of semianalytical triangulation. 
The first consideration concerns the adjustment of residual parallaxes. It 
is evident that, once the model has been formed in the best instrumental way, 
some residual parallaxes can be found. If the instrument is provided with a paral 
laxes meter or if, as in the Stereosimplex, it is possible to mount a suitable device, 
with which before reading the instrumental coordinates, the operator can measure 
the observed parallaxes. Formulas can be used so that, with the residual parallaxes, 
read on several points, we can compute the most probable correction of the orien 
tation parameters and consequently the corrections of the read coordinates of the 
model points. This technique has not yet been proved, but on this subject many 
photogrammetrists have written papers. As soon as possible, we intend to perform 
some experiments in this sense. 
We want essentially to see whether it is possible to avoid the work of a ca 
reful elimination of the parallaxes to the operator, by means of the instrumental 
orientation parameters. 
This method would become more complicated as regards to the computations, 
but it would gain in easiness and in swiftness from the instrumental point of view 
and should allow a greater precision in the model determination. Other possibili 
ties of semianalytical triangulation concern the correction of the distortion, of the 
atmosferic refraction, of the film shrinkage, and so on. Of course, in the case of semi 
analytical triangulation it is possible to correct the effects produced by these errors 
on the coordinates of the model. 
Another computation program has been studied for computation of the absolute 
orientation of the strip in the case the first model is not oriented ; this computation is 
practically equal to that for the models connection ; however, this program takes into 
account the earth curvature for the correction of the heights of the surveyed points. 
The computation program used for the adjustment of the semianalytical 
strips is the same used for the analogical and analytical triangulations executed 
in our Institute. 
The results of the experiments, in fact, have shown systematic errors of the 
same kind of those we found in the analogical and analytical strips ; we must say, 
however, that, at present, the adjustment program is not very important, since 
it is used only to make the results obtained by semianalytical triangulation compara 
ble with those obtained by other methods, and to prove the reliability of the method.