“…Therefore, it is most important that these distributions be accurate at their right tail than elsewhere. It is well known, however, that statistics of extreme values are the most affected by uncertainties, even though ad hoc methods have been proposed to handle extreme values (e.g., Roberts, 1979;Williams, 1984;Drufuca and Giuliano, 1978;Horowitz and Barakat, 1979;Chock and Sluchak, 1986;and Surman et al, 1987). It is unfortunate, from a statistical point of view, that U.S. air quality standards were chosen as values that can be exceeded only once a year.…”