Published online by Cambridge University Press: 03 August 2017
Polar motion data for the period 1981–1985 are used to obtain a combined solution from Doppler, Satellite Laser Ranging, and Astrometric observations. The combined solution is a weighted average of the three series, with weights determined from reported errors which are scaled so that they agree with errors estimated from differences among the various series. The combined solution is effective in removing spurious deviations in the pole path which appear in a single series. However, we also show that estimated errors can be unreliable when derived from short time series, when one series is much less noisy than the others. Thus, a combined solution where weights depend upon estimated errors can yield poor results, and we demonstrate this effect by comparing a combined solution for 1984–85 with the independent IRIS series.