Intermediate-term earthquake prediction

The problems in predicting earthquakes have been attacked by phenomenological methods from pre-historic times to the present. The associations of presumed precursors with large earthquakes often have been remarked upon. the difficulty in identifying whether such correlations are due to some chance coincidence or are real precursors is that usually one notes the associations only in the relatively short time intervals before the large events. Only rarely, if ever, is notice taken of whether the presumed precursor is to be found in the rather long intervals that follow large earthquakes, or in fact is absent in these post-earthquake intervals. If there are enough examples, the presumed correlation fails as a precursor in the former case, while in the latter case the precursor would be verified. Unfortunately, the observer is usually not concerned with the 'uniteresting' intervals that have no large earthquakes.