The results of the electrical breakdown time-delay as a function of the relaxation time τ, known as ‘the memory curve’, are presented. The memory curve was experimentally established for the relaxation times from 3 ms to 300 s, for the Ne-filled diode at 7 mbar pressure. Statistical analysis was used to analyse the obtained time-delay values. The results show that for the relaxation times τ ⩽ 0.1 s the dominant part of the total time-delay is the breakdown formative time (tf ∼ 0.1 ms). The breakdown formative time obtained in this investigation shows the same ‘memory effect’ as the breakdown total time-delay, because it slowly increases with τ. For τ > 0.1 s, the statistical time-delay makes up the dominant part of the total time-delay, as was reported earlier.

The processes responsible for initiation of electrical breakdown in xenon-filled tube with two spherical iron electrodes at 2.7-mbar pressure have been analyzed. The analysis is based on the experimental data of electrical breakdown time delay as a function of afterglow period. It is shown that positive ions remaining from previous discharge, as well as positive ions created in mutual collisions of metastable atoms in afterglow, have a dominant role in secondary emission of electrons from the cathode which lead to initiation of breakdown in early afterglow. In late afterglow, dominant role in initiation of breakdown is taken by N(4S) atoms formed during the discharge by dissociation of ground state nitrogen molecules that are present as impurities in xenon. When the concentration of N(4S) atoms decreases sufficiently, the initiation of breakdown is caused by cosmic radiation. Small doses of gamma-ray irradiation also contribute to the initiation of breakdown, but only for large values of the afterglow period.