Correction for Universal non-normality

It may be asked if testing for normality and, when necessary, correction for universal non-normality is worth the trouble. To answer this question it is desirable to have regard to the logical position of the statistician, concerned with drawing inferences from samples whose characteristic approach may be defined as reductio ad paene absurdum: if an event is highly improbable it must be regarded for practical purposes as impossible.

Page 239