A prescription for the error analysis of experimental data in the case of stochastic background is formulated. Several relations are given which allow to establish the significance of mother-daughter relationships obtained from delayed coincidences. Both, the probability that a cascade is produced randomly and the probability that the parameters of an observed event chain are incompatible with known properties of a given species are formulated. The expressions given are applicable also in cases of poor statistics down even to single events.

No abstract