Fluctuation theorem and natural time analysis

Upon employing a natural time window of fixed length sliding through a time series, an explicit interrelation between the variability $\beta$ of the variance $\kappa_1$($=< \chi^2 > - < \chi >^2$) of natural time $\chi$ and events' correlations is obtained. In addition, we investigate the application of the fluctuation theorem, which is a general result for systems far from equilibrium, to the variability $\beta$. We consider for example, major earthquakes that are nonequilibrium critical phenomena. We find that four (out of five) mainshocks in California during 1979-2003 were preceded by $\beta$ minima lower than the relative thresholds deduced from the fluctuation theorem, thus signalling an impending major event.