Extensive data on fibre properties are analysed for variability, both Type I and Type II errors being taken into account. Real-difference values between sample means, the order of Type II errors in detecting differences that may be considered technologically significant, and the number of tests required are presented in tables.