In studies of nitrate leaching both experimenters and modellers experi
ence problems arising from soil variability. Because of the small-scal
e heterogeneity that gives rise to mobile and immobile categories of w
ater, both measurements and modelling are easiest in homogeneous sandy
soils and most difficult in strongly structured clay soils. There are
also parallels at plot and field scale in the problems caused to expe
rimenters by log-normal distributions of nitrate concentrations and th
ose caused to modellers by non-linearity in models. All researchers ne
ed to be aware that a reliable estimate of the mean from a set of meas
urements or a model may necessitate considerations of variances as wel
l as means.