In a previous paper (Carter, 1981) I suggested that the Chandler component of polar motion may be frequency modulated, as a function of the instantaneous polar motion magnitude. A composite of observational records compiled by S. C. Chandler and the International Latitude Service series suggests a variation in the magnitude at a frequency of approximately 0.0118 cycles per year (cpy). The spectrum produced by the resulting frequency modulation would depend on the modulation index. Comparison of spectra synthesized various degrees of frequency modulation and those derived from observational data, suggest a frequency increase of 0.15 to 0.30 cpy/arc sec decrease in the polar motion magnitude. This degree of frequency modulation would cause 'splitting' of the Chandler component, producing dual spectral peaks at periods of 427 and 439 days, similar to that reported by several investigators beginning with Chandler (1901a, b).