176 E. LANCASTER JONES 
problem of least-square adjustment must be governed by considera- 
tions of the accuracy justified in the particular survey, and the most 
economical method of executing the calculations involved. Adjust- 
ment involving elimination of the correlates of the first-remove cells 
on the lines of §§4 and 5, using equations (6) to (11), is a simple, 
reasonably rapid process, readily applicable to any cell in any net- 
work however complicated. When the network is homogeneous, i.e., 
when all the cells are polygons of the same species and regular in ar- 
rangement, the more rapid approximations discussed in §§6 and 7 
may be used with advantage. Also when there are outstanding runs 
or chains of cells equations (16) to (19) are easy to apply and give 
very rapid approximations. 

Fic. 7 
For general purposes, the procedure of §§4 and 5 is recommended 
as being simple, universally applicable and almost as rapid as any of 
the subsequent processes when account is taken of all the stages of 
the operation. Asa test of this procedure, an adjustment made fora 
network of 56 Eétvés gravity stations may be cited. The observations 
were made in the course of a practical survey in Cumberland. The 
stations are shown as open circles in Figure 7, where they are connected 
by broken-line links to form 80 triangles numbered 1 to 85, numbers 
26 to 30 being omitted. It is evident that the network thus formed is 
typical of a large regional survey and that the procedure of linking 
could be adapted to any group of stations, however numerous. From 
the measured data of the gravity survey the increments in g were 
calculated for every link, and from these the excesses were derived 
for each triangle of the network. The excesses for the first ten triangles 
are tabulated in Table VIII in convenient units which were subse- 
836 
