NMR data for some water-saturated sandstones show distributions of relaxation times covering ranges of a thousand or more. Pore size distributions have been associated with distributions of relaxation times and can also cover wide ranges. Several dozen good sets of NMR relaxation data for water in porous sandstones have been analyzed in terms of continuous distributions of exponential components. About a dozen of these are for sandstones having significant relaxation time components over ranges of factors of a thousand. In all cases adequately good fits to the data could be obtained with distributions of relaxation times that were monomodal when plotted as functions of log-time (or log-rate). Thus, it appears that bimodality (or multimodality) for the logarithmic plots is not demanded by these particular sets of data, although these distributions plotted linearly are not monomodal. On the other hand, many multimodal distributions can always be found giving adequate fits to the data, since excessively sharp detail is not resolvable. Many programs using regularization methods to prevent excessive detail in computed distributions tend to give undershoot at sides of peaks, and noise tends to give not-quite-periodic oscillations. Lack of adequate range and density of either data points or computed points can lead to multimodal computed solutions. Some resolution expressions are used to indicate what level of detail in a computed distribution is meaningful for a given data set.
