In small confined systems predictions for the melting point strongly depend on the choice of quantity and on the way it is computed, even yielding divergent and ambiguous results. We present a very simple quantity that allows us to control these problems-the variance of the block averaged interparticle distance fluctuations.