NM RN = I —————————————— 
2. Stereoblocks 
The unit for adjustment will be termed a "stereoblock", 
which, to fix ideas, may be taken as a set of three adjacent 
strips of three photographs, although these dimensions may 
be varied at will. 
Each stereoblock will be adjusted as a single unit, by 
the method outlined earlier, each "observed quantity" for 
the internal orientation being, in the usual Church manner, 
the space angle subtended at the camera centre (0,0,-f) by 
two photo points (x,y,0), (x',y',0), and the corresponding 
approximate coordinates in space of the camera station and 
ground points being (U,V,W), (X,Y,Z), (X',Y',Z'). 
The form of the coefficients and constant term for the 
observation equations, derived in Appendix B, is thought to 
be new; since they depend purely on scalar and vector 
products of position vectors, they seem particularly suited 
to electronic computing, In addition to these, linear, 
coefficients the derivation of the second, and if necessary 
higher, order terms is also given in the same appendix, 
although these are not of course needed for the observation 
equations, 
If there are just p photo points on a photograph, then it 
is easily seen that there are just (2p-5) independent obser- 
vation equations corresponding to them, and care must be 
taken that the (2p-3) pairs of photo points chosen to form 
these equations do in fact give independent conditions, 
Again, since it is at the stage of forming the observation 
equations that any bad observations show up, it is essential 
to avoid a complicated criss-crossing choice of pairs, since 
to reject a point would upset the whole scheme. For this 
reason, it is proposed to include the principal point as an 
additional unidentifiedx photo point, the corresponding 
ground point being taken as (S,T,0) in the datum plane (220); 
of the (2p-1) pairs which now need to be chosen, p pairs can 
be taken to be the principal point paired with each of the 
true photo points in turn, while the remaining (p-1) pairs 
are obtained by arranging the p points in order, and pairing 
each point with the one which follows it (if any). Rejection 
of a point now merely involves deleting three old observation 
equations and including one new one ("leap-frogging" over the 
rejected point); further details and numerical examples are 
given in Appendix A. 
In addition to these internal orientation conditions, 
any external conditions on points of the block may simultan- 
eously be applied in the observation equations; thus if a point 
is a point of ground control, planimetric, altimetric or both,