A deconvolution is the process of removing the influence of extraneous variation from an apparent cumulative distribution. Extraneous variation--such as random errors in measurement-- has the effect of inflating observed variation relative to true population variation. The cumulative distribution that will be estimated when extraneous variation is present is the convolution of the population distribution (which is the cumulative distribution of interest) and the distribution of the extraneous variable. The convolution cumulative distribution will be flatter (have longer tails) than the population cumulative distribution.