DETERMINATION BY TORSION BALANCE 79 
squares of the respective “probable errors” of the two pendulum 
observations and of the two determinations of the elevation. The 
“probable error” of the pendulum determination of gravity at each 
key station can be decreased by reoccupation of that station or by 
the occupation of additional closely adjacent stations which are tied 
together by torsion balance determinations of the value of Ag be- 
tween the key station and each of the subsidiary stations. The de- 
crease in the “‘probable error’”’ will vary as the reciprocal of the square 
root of the number of stations. 
The minimum interval between pendulum observations which 
are to supplement ordinary, good torsion balance surveys in good, 
fair, or even slightly poor torsion balance terrane is given in Table V. 
TABLE V 
MINIMUM INTERVAL AT WHICH PENDULUM OBSERVATIONS SHOULD BE USED TO TORSION 
BALANCE SURVEYS 
“Probable Error” ‘Probable Error” of Pendulum Determination of Ag 
of Torsion Bal- in 1X10* Dynes 
lance Determina- 
tion of Agin 10-4 
Dynes for Places +10 +5 +23 +1 
10 Km. A part 
Airline Distance Interval in Kilometers 
A* B A B A B A B A B 
Poorer ‘‘Good”’ 
surveys...... 7 & 40 75 Io 6.20 2205 0.3 0.5 
Average ‘‘Good” 
surveyS...... A 8 120 220 30.0 «55 8 14 I.0 1.5 
Better “‘Good”’ 
suIVeyS...... 2 «If 500 650 120 210 30 55 4-5 6 
* A. The two places are connected by a single direct or two indirect torsion balance traverses. 
B. The two places are connected by several traverses. 
In surveys primarily by the pendulum, or in pendulum surveys which 
are supplemented by sketchy torsion-balance surveys, the intervals 
between the pendulum stations very much less than the minimum 
intervals of Table IV may be used in order to increase the accuracy 
of the survey. 
409 
