It is shown that the location of the set of m + 1 observations with mi
nimal diameter, within local data, is a robust estimator of the locati
on of a vertical tangent in a distribution function. The rate of consi
stency of these estimators is shown to be the same as that of asymptot
ically efficient estimators for the same model. Robustness means (1) o
nly properties of the distribution local to the vertical tangent play
a role in the asymptotics, and (2) these asymptotics can be proven giv
en approximate information about just two parameters, the shape and qu
antile of the vertical tangent.