Previous investigators, when measuring the electric strength of hydrocarbon liquids with short-duration rectangular pulses, have assumed that the statistical component of the breakdown time was insignificant com pared with the formative time. In the present investigation, however, the time to breakdown was measured directly by the use of step-function pulses, and clear evidence for a statistical time lag was found. The formative time was ~ 0.1 us, being less than that given by previous estimates. A statistical interpretation of short-pulse measurements is presented and this provides a consistent explanation of the results of other workers. Furthermore, by using an experimentally derived equation for the variation of the mean rate of breakdown
f
(
E
) with applied stress
E
, it has been shown that the form of the relationship between strength and pulse duration obtained by other workers agrees with that obtained by a statistical analysis. Experiments on air-saturated
n
-hexane with both short-duration and step-function pulses support the statistical ideas presented and indicate that electrode conditions are extremely important. It was found that strength and time to breakdown were affected by the number of breakdown measurements on a sample. Experiments with gas-free
n
-hexane and non-uniform fields have demonstrated the importance of air content when long duration pulses are used. It was found that, although the statistical time lag was insignificant, formative time lags as long as 10 ys occurred with a point cathode-sphere anode configuration.
Explanation of the memory effect in argon
