74 DONALD C. BARTON 
of the length of the traverse. As SJ-pey., somewhat commonly in 
practice is held approximately constant, the “probable error” of the 
determination of Ag by ordinary good torsion balance surveys can 
be expressed in terms of Ag per some standard length of traverse. 
That this is so, can be seen by analysis of the error of closure of 
those 45 traverses. The error of closure of each closure of Ag in each 
of those traverses can be scaled up or down by the formula 
(9) EC = io 
to the error of closure (ECio), which the traverse would have had, if 
its length had been 10 kilometers instead of (L) kilometers, and if 
there had been no change in the station interval. The resulting errors 
of closure per 1o-kilometer traverses have been sorted by length of 
the station interval and are given in Table IV. The respective errors 
of closure and the respective “probable errors” of Ag for the traverses 
of 100, 200, 300, 400, 500, 600 meters station interval, can be seen to 
TABLE IV 
ERRORS OF CLOSURE SCALED TO A 10 KM; LENGTH OF THE TRAVERSE 
Station interval in meters 100 200 300 400 500 600 700 800 
Error of closureofAg 8.7 6.8 15.6 4.1 8.2 Fok 4-3 6.0 
in 10 4 dynes 4-9 6.8 5-0 5-0 Fo] 6.6 gas 
Se4) | e4-SheS-Onva4-7) 9 Oso a a9 
2237 Ged) 14-2 S-4) eas) a 
2.0 3-2 1.9 4-4 4.0 
P35) 1.6 3-9 1.8 
O:0) a: 3-9 0.0 
3.6 0.0 
3-3 
1.6 
1.4 
“Probable error” of 
LEAVCISC =e rae ee erent) Seigéy) Ser lay Gell acStoZl arent 
“Probable error” for all 45 traverses: +3.9 
be independent of the length of the station interval. That “probable 
error” ranges only from 0.32 milledyne for the traverses with a station 
interval of approximately 600 meters to 0.46 milledyne for the traverses 
of approximately 300 meters; and for the traverses of approximately 
100, 200, 500, and 600 meters length, that ‘‘probable error” ranges 
only from 0.32 to 0.36 milledyne. The “‘probable error” of the indi- 
vidual observations, however, was considerably larger in the traverses 
404 
