A new method to approach earth noise field problems in inertial component testing is presented. This method is considered superior to accepted 'quiet site' and 'motion suppression structure' techniques. The approach is to describe earth noise by determining: (1) the class of the random motion; (2) the level of motion; (3) the mode of motion; and (4) the component relationships. The efficacy and rationale of this technique is demonstrated in an experiment that results in the removal of earth noise error terms from gyro data.