A comparison of a revised ILS polar motion series for the years 1901 to 1970 with an atmospheric time series shows evidence for coherence near the period of the 14 month Chandler motion. The same ILS series, and two other BIH series, have also been used to compute shifts in the apparent pole position near the times of the 1960 Chilean and 1964 Alaskan earthquakes. These shifts are not convincingly similar to those predicted by other investigators from earthquake source parameters. In addition, there is disagreement among the shifts computed from the various polar motion data sets. Thus, a conclusive determination of the relative roles of atmospheric, earthquake and other sources in the excitation of polar motion will probably require improved observations of both the proposed excitation phenomena and the earth's polar motion.