The paper gives some specific instances of the wide variations which can occur in the measured electric strength of industrial insulating materials, arising partly within the material itself and partly on account of differences in the test conditions. Some of the results are subjected to an elementary statistical analysis.It is shown that short-time tests may be quite misleading as an indication of the relative merits in service of materials of different kinds, and it is concluded that the performance of a material can be found only by trial over a prolonged period. The paper includes the results of electric-strength tests at 50 c/s for periods up to two years, and these are compared with the results of tests lasting for only a few minutes.