Occurrence of destructive earthquakes in Tokyo area during historical time has been investigated by several authors. Possibility of 69-year as well as 36-year periodicity of earthquake occurrence was proposed on the basis of periodgram analyses. A basis of these hypotheses is Schuster's criterion which is applicable only to any one harmonic term obtained from a random sampling. Although the above periodicities are obtained by applying the criterion to the largest of the harmonic terms, the writer believes that, in order to test the largest, we must adopt Fisher's test.
A periodgram of a discrete time series of the earthquake frequency is computed for the Tokyo area. It is shown that the periodicity corresponding to its greatest peak is not statistically significant by Fisher's test at significance level 0.1.