The evaluation of these data allows cautiosly to observe that the corre- 
lation between errors of ground control and plotting errors is conside- 
rably variable from an operator to the other. Of course the operator's 
technical experience and age must be particularly important in the error 
adjustment during the formation and absolute orientation of photogram- 
metric models. 
Generally, to a low values of the correlation coefficient and plotting errors 
should correspond a greater skill of the operator in the realization of 
the compromise between relative orientation and absolute orientation of 
the stereograms and in the performance of the following model plotting. 
Unfortunately the tests of correlations are resulted no significants. The 
number of samples is very small and has not allowed to investigate care- 
fully through an analysis of covariance including a test for the parallelism 
of the separate regression lines [5]. 
5 
Without any doubt, the experimental research started by Mr. Le Divelec 
should deserve to be continued and extended increasing the number of 
observations and taking into consideration the error influence on the 
control points in the plotting of a group of adjacent models. The further 
information which might be obtained, should, certainly, supply a much 
more important contribution to the knowledge of the different plotting 
problems. 
From the information obtained through this pilot research, we are able to 
confirm that the errors in the establishment of ground control points are 
changing both the relative and absolute orientation of stereograms. Being 
impossible to separate the wrong control points, the formation of the 
photogrammetric models is in fact obtained through a compromise bet- 
ween residual y-parallaxes and residual errors on ground control. 
Besides, through further investigations, it should be possible, from the 
examination of the obtained results, to evaluate in a more convenient 
way also the errors in the measurements of transversal parallaxes, chara- 
cteristical of the plotter used, taking into account that the reading of the 
residual parallaxes may sometimes be difficult [6]. 
Generally speaking from the different information obtained in the research 
reported in this paper we can confirm that the total accuracy obtained, 
comparing the numerical values of the machine coordinates of the plotted