The Stringer bound is a widely used nonparametric 100(1 - alpha)% uppe
r confidence bound for the fraction of errors in an accounting populat
ion. This bound has been found in practice to be rather conservative.
In the present paper, we give recursive relations for obtaining the ex
act distribution of the Stringer bound in the case where the underlyin
g distribution of the taintings is a uniform distribution on the inter
val [0,1], or a distribution with positive mass at aero and conditiona
lly uniform on (0,1]. Based on these recurrence relations, we find a c
oncrete counterexample which shows that the Stringer bound is not alwa
ys conservative.