A cubic spline curve fit to histograms of grain-size data from pyroclastic deposits affords a rapid means to distinguish subpopulations in complex size distributions. Using interactive microcomputer graphics, the operator can isolate and quantitatively characterize sub-populations by a method of successive Gaussian subtractions. Percentile measures of the entire size range are not needed, and segments of the whole size distribution can be analyzed independently. Recurrent subpopulations with similar means and standard deviations are consistent among the more than 140 samples of pyroclastic-surge deposits from Sugarloaf in Arizona and Vulcano in Italy. The entire data set can be replicated by synthetic curves consisting of various proportions of five identified Gaussian components, each of which we attribute to a specific mechanism of transport and deposition.