Hydrogen usually occurs in sheared silicate rocks and its concentration much fluctuates, spatially and temporally. The concentration of H2 from activee faults associated with historical earthquakes usually amounts to as high as several percent in maximum. On the other hand, the H2 concentration from Quaternary faults not associated with historical earthquakes is at most 100ppm. Gases from both types of active faults are impoverished in O2. A series of laboratory experiments showed that (1) hydrogen is produced from a mixture of fresh rock poweder and water and (2) the production rate fluctuates according to rock type, reaction temperature and atmosphere above the mixture. The mixture made of pegmatite powder gave H2 as high as several percent. Oxygen decreased when the reaction proceeded. The experiments suggests that the fresh mineral surface formed by tectonic stresses reacts with groundwater to produce H2. Since the mineral surface loses its activity with time, discrimination between recently moved faults and other Quaternary faults can be made by the H2 concentration. Hydrogen isotope thermomenter, as well as field evidence, suggests a deepseated origin of H2 in an active fault. Successive measurements showed that much H2 simultaneously issued at several monitoring stations on historical active faults and that H2 appeared in bubble gases from mineral springs during about one month prior to earthquakes. The evidence suggests that H2 measurement at monitoring stations gives information at depth on mechanisms that operate perior to earthquakes.
