When the probability distribution of measurements is unknown, repetitive random resampling with replacement of the available data set offers a close approximation to the true distribution, and consequently, an estimate of parameters of the population from which the data set is considered to be a sample. The approach is illustrated by means of lead contamination in an aquifer, where a small set of measurements using ion-sensitive electrodes provides inferences about the expectation and the standard deviation of the Pb2+ ion concentration. (C) 2008 Elsevier B.V. All rights reserved.