On the Test of a Polyester Aerial Photographic Film T-008 
109 
PGR 
I T r 
s r 
Fig. 8. Way of taking the runniug mean. 
cause the observation errors occur arbitrary. Running means of 28 points 
were calculated by following formulae. (Fig. 8.) 
A=^(p a +p b +p c +p d +p e ) 
P=j(Pp+P q +P s ) 
Q=j(p P +p q +Pr+Pt) 
With these running means, the closing error V and the mean square error <r are 
calculated, and contour lines of 5 p interval are drawn for each stereo-model. 
(Some examples are shown on Fig. 10.) 
3. Investigation of the results 
1) Results of the calculation of V, V' etc. for all models are shown on 
Table 3. In spite of most observations were performed with diaposi- 
tives printed on CNT film of polyester base (Fujilith Orthofilm), Toyo 
Koku Jigyo Co., Ltd. observed with original negatives. 
For the comparison to TAC film, results of the observation executed 
by Toyo Koku Jigyo Co., Ltd. are shown on Table 4. In this case, the 
original negatives were taken with Super XX and as diapositives 
CNT films were used. 
2) Re-Observation 
Because of closing errors V, V' of No. (11) and No. (12) stereo-models 
(1/30,000) were found extremely large, they were observed again. But 
by the investigation of the primary observations it was concluded that 
there must be some trouble by the printing process of diapositives. 
So we applied various combinations of the printer and the plotting 
instrument for the re-observation on Table 5. No. (38) and No. (43) 
are the re-observation of the same diapositives used by the primary 
observation. 
4. Some analysis of the results 
1) Frequency distribution and the tolerances of V, V' and a-. The values 
of V, V' and o- are thought to vary with the photo-scale, plotters and their ad 
justments, the technique of operators etc. . The results of the experiment are 
shown on Table 6 classified according to these elements, and Fig. 9 shows the