I recall a textbook example of this. Early in the 1600's a scientific cohort of
Galilelo was doing some early experiments on magnetism with a hanging
piece of lodestone (from a string). He (I don't recall his name, but I don't
think it was Galileo himself) reported that one afternoon he was doing his
experiments and suddenly the lodestone lost all its magnetic properties.
A few minutes later a strong earthquake struck. He and many since have
been convinced earthquakes could be predicted using this observation, but
it has not yet happened. Maybe not enough observations in enough
places yet?