No observations are absolutely trustworthy.
Rejection of Outliers, Technometrics, Volume 2, Number 2, May 1960 (p. 124)
