The frequency estimation errors due to both techniques are compared, and it is shown that there are some interesting similarities and differences. A major result is that for even relatively short data lengths the maximum entropy approach should not automatically be regarded as the superior technique for the frequency estimation of truncated real sinusoids.