Abstract
Statistical tests for the Poisson process, the magnitude distribution and the distribution of the annual maximum magnitude of earthquake occurrences in and near Japan listed in JMA catalogue from 1885 to 1974 were examined. The goodness of fit tests for the time occurrences of large earthquakes show that the Poisson process is rejected in some unit areas of approximately 10000km^2 as shown in Figs. 1 and 2 since earthquakes tend to come in groups. Except for a few unit areas, the goodness of fit tests show that the truncated magnitude distribution derived from the Gutenberg and Richter's law is accepted as shown in Figs. 4 and 5. Results of the Kolmogorov-Smirnov goodness of fit tests for the annual maximum magnitude indicate that the extreme value distribution derived from the Poisson process and the magnitude distribution is adequate to explain the observed annual maximum magnitude in and near Japan as shown in Figs. 7 and 8.
