valleys insofar as practicable). Though of secondary interest, vertical 
densification was also to be accomplished. 
As previously reported (Brown (1973),(1974)) the error model 
employed at the time in the DBA program Combat II contained 29 terms and 
was of the form: 
Ax 
IN 
= Q,x td,y t dx? t a,xy t asy? cau x^y c a,xy? 
+2 (1X t eaxy * cay? * cox t eai y t eux? t ry) 
| 4x (K,r* + Kar* + Kr?) 
+ P, (PP +3x2) + 2P, xy 
* 6x, +(%) be i- d | | 
byx + bay + bax® + b,xy + bay? + hgx®y + b,xy? 
+L (c, x +exy + cs tex texy * G xy? * cy?) 
+y (Kor? * K,r* * Kr?) 
* 2P, xy * P 6? € 3y?) 
Oy, + (t óc. 
in which the various coefficients may be interpreted in groups to correspond ) 
to the following: 
(a) 
(b) 
(c) 
(d) 
(e) 
41,055,..., 0 
1? 23?***9 7 
C1, C555. , C, 
K,,K,,K;: 
Pi) P,: 
Xp, 9p, cs 
empirical coefficients defining combination of 
mean uncompensated film deformation and 
anomalous distortion; 
coefficients defining curvature or unflatness 
of platen; 
coefficients of radial distortion; 
coefficients of decentering distortion; 
biases in enforced. elements of interior 
orientation. 
Being a 'raw' mixed empirical and analytical model, the above formulation 
can be expected to suffer from strong correlations between various parameters. 
However, as the writer explained in discussions at the 1974 Stuttgart 
«16-