Accuracy of vertical deflection determination by present-day inertial instrumentation

An analysis of results obtained in the Canadian Rock Mountains indicates that the observation of deflection differences along the same line can be repeated with a precision of about 0.5 sec but that there are systematic discrepancies between the forward and the backward running of the same line. A comparison with the available astronomically determined deflections also shows systematic differences of 2 sec and 3 sec. These errors are most likely due to the overshooting of the Kalman procedure at gradient changes. It appears that the software can be altered in such a way that deflection differences between stations, not more than half an hour of travel time apart, can be determined by the inertial system with an accuracy of better than + or - 1 sec.