Possible Precision of Statistical Insulation Test Methods

The report deals with insulation test methods where voltages of constant shape but different amplitudes are applied to a test object to determine the voltage which corresponds to a certain probability of flashover. A hypothetical test method is considered which gives the minimum statistical error in the determination of that voltage for any given number of voltage applications. It is shown that some existinig test methods give nearly this result. There is therefore no need for further development of such methods. Present methods to determine voltages corresponding to low probability of flashover are criticized.