“…To avoid confusion it is convenient to subtract from (1) the expected values of the probabilities, (PF> and (P n > and write in the form x (2) X can be regarded as a sum of steps xn with random lengths, and one can study the statistics for an R-C interval for X (i.e., system p,robability of failure, in terms of the statistics for R-C intervals of the x n ' i.e., flaw state probabilities).…”