It is found that the mountain and friction torques contribute little to the changes of the autocovariance function of the equatorial angular momentum with lag. It is almost exclusively the earth's rotation and its shape which induce these changes. Mass and wind terms are hardly correlated in contrast to predictions of normal mode theories. However there is a strong correlation of mountain torques and wind terms.
As for the axial component it is found that the autocovariance of angular momentum decays rather slowly. The crosscovariance of mountain and friction torques does hardly contribute to the changes of this autocovariance with lag. However the data quality is not sufficiantly high to explain the observed slow decay of the autocovariance in terms of the autocovariances of the torques.