Abstract :
The application of statistical methods to an examination of the value of current specifications in determining the mechanical suitability of line insulators shows that unexpectedly great aberrations may occur. A detailed analysis of a Central Electricity Board specification, for instance, illustrates how appreciable, are the chances of passing a ba, tch, containing a large number of defective units, and it may be argued that too much importance is attached to a test of doubtful value. It may be possible to go some way towards reducing the discrepancy limits by adopting a more actuarial attitude towards the interpretation of results. Specification levels might be arrived at by the methods described in B. S. S. 600¿1935, and a hypothetical case has been examined. In practice it would first be necessary to determine by research whether (1) The frequency curve for line units was statistically determinate. (2) Reliable specification levels could be determined. It is also pointed out that samples of not less than 30¿50 units should be tested, although these might be drawn from larger batches. The final section of the paper contains the author´s conclusions as to the value of mechanical consistency and a suggestion that resistance to thermal stresses might profitably receive more attention.
