DETERMINATION BY TORSION BALANCE 77 
servations, and, therefore, the cost and tediousness of too close a net 
of stations puts a practical limit to the maintenance of that accuracy 
of +0.35 milledyne between places 1o kilometers apart, airline dis- 
tance, or for the maintenance of any particular grade of accuracy. In 
reconnaissance in oil work, a station interval less than 250 meters 
commonly is impracticable, and if the “‘probable error” of the in- 
dividual gradient observations rises above +2 EF, it becomes imprac- 
ticable to maintain an accuracy of +0.35 milledyne in the determina- 
tion of relative gravity between places approximately 1o kilometers 
apart, airline distance. In some areas in South Texas, in which the 
caliche lies close to the surface, the ‘‘probable error”’ of individual ob- 
servations probably is greater than +7 EH; and in certain areas in 
East Texas, the ‘‘probable error” of individual observations probably 
is greater than +5 EL. The “probable error’’ of the determination of 
gravity between two places approximately ro kilometers apart in 
such areas will be very much greater than +0.35 milledyne. 
Knowledge of the ‘‘probable error’’ of the determination of Ag 
by torsion balance surveys is important in certain types of combina- 
tions of torsion balance and pendulum surveys. According to a rather 
common suggestion from “pure” geophysicists, pendulum observa- 
tions of relative gravity at a net of key stations should be used to 
provide a framework of accurate gravity benchmarks to which the 
supposedly less accurate torsion balance surveys would be tied and 
adjusted. But the accuracy of the torsion balance surveys will be in- 
creased thereby only if the “‘probable error’’ of the determination of 
relative gravity by the pendulum between two stations is less than 
that by the torsion balance surveys. The determination of relative 
gravity by the relative pendulum is independent of the distance be- 
tween stations; the value of Ag between two stations is simply the 
difference between the observed values of relative gravity at the two 
stations, and the “‘probable error” of Ag between the two stations is 
the square root of the sum of the squares of the respective ‘“‘probable 
errors” of the individual observations. The “‘probable error” of the 
determination of Ag between two places by good torsion balance 
surveys varies, in practice, with the square root of the distance be- 
tween the two places. The torsion balance surveys will measure Ag 
more accurately than the pendulum if the two stations can be con- 
nected by very short torsion balance traverses, and the pendulum 
will measure Ag more accurately than the torsion balance if the two 
stations can be connected only by very long torsion balance traverses. 
407 
