The homogeneous set of 80-year-long (1900-1979) International Latitude Service (ILS) polar motion data is analyzed using the autoregressive method (Chao and Gilbert, 1980) which resolves and produces estimates for the complex frequency (or frequency and Q) and complex amplitude (or amplitude and phase) of each harmonic component in the data. Principal conclusion of this analysis are that (1) the ILS data support the multiple-component hypothesis of the Chandler wobble (it is found that the Chandler wobble can be adequately modeled as a linear combination of four (coherent) harmonic components, each of which represents a steady, nearly circular, prograte motion, a behavior that is inconsistent with the hypothesis of a single Chandler period excited in a temporally and/or spatially random fashion). (2) the four-component Chandler wobble model ''explains'' the apparent phase reversal during 1920-1940 and the pre-1950 empirical period-amplitude relation, (3) the annual wobble is shown to be rather stationary over the years both in amplitude and in phase and no evidence is found to support the large variations reported by earlier investigations. (4) the Markowitz wobble is found to support the large variations reported by earlier investigations. (4) the Markowitz wobble is found to be marginally retrograde and appears to have a complicated behavior which cannot be resolved because of the shortness of the data set.