networks or blocks so that the undetected model errors do not lead to 
any serious destortions of the final coordinates of the points. 
2. THE ADJUSTMENT MODEL 
: ; à + ; ; ; j 
Having n observations x’, i-l, ...,n, n independent relationships, in 
general non-linear, can be formulated. 
= FL. xbesdadndrilge. wn (1.1) 
with the inverse relationships of the form 
) (1,2) 
The mean values SK of the quantities y" are in general unknown, but a 
measuring procedure can be designed in such a way that for a subset of 
the y. the mean values are known and are equal to zero. 
In view of this, the relationships (1.1) can be separated into two 
subsets. 
ya y* s x* 2) a,8 = 1, ><. , m 
= with m+b=n (1.3) 
y y. x" ) Bat = Le e: D | 
with 
In order to derive a consistent set of linear relationships approximate 
values are assumed. 
Y a Y and X, = X 
Omitting second and higher order terms, Taylor’s expansion applied in 
(1.3) gives 
a a 
ay, ui 
i 
= A . 
p p ES We 
Ay u. 
= 0 i 
with 
k ; 
ut = N : and Axt -oxloxi 
i act xd k - 0 - oO 
9^3, A BR 
Ys FX YS 
similarly the inverse relationship (1.2) gives 
- 94 -