The nature of the irregular fluctuations in the speed of the Earth's rotation was investigated using ninety-day means of UT2 - A.1 determined at the U. S. Naval Observatory. Data from June 1955 to April 1978 were included in this study. Statistical analysis of the excess length of day shows no evidence for the persistence of discrete values for periods on the order of five years. No statistical basis for the existence of discrete “turning points” in the rotational speed could be found.
Spectral analysis of the acceleration data shows that the rotational acceleration of the Earth during this period of time may be represented by a constant term plus random changes in acceleration occurring with a frequency greater than once per year. The magnitude of these changes appear to be consistent with estimates of meteorologically induced changes in the rotational acceleration.
An autoregressive integrated moving average (ARIMA) model was fit to the excess length of day data. This model permits simulated series of excess length of day data to be constructed. These simulated series show a statistical similarity to observations made since 1820. However the apparently large changes in the acceleration which occurred around 1870 and 1900 are twice that which can be reasonably accounted for by this model. The details of this analysis will be published later.