The choice of analytical method and sample pretreatment has significant consequences for the shape of particle-size distribution (PSD) frequency curves, and therefore for the evaluation of soil textural parameters. Here, the comparison of several granulometric methods based on different physical principles is presented: wet sieving (WS), based on gravity and the mechanical force of water, was combined with pipette (PP) sedimentation/settling velocity, Micromeritics SediGraph (MS) sedimentation/x-ray attenuation and Coulter Counter (CC) electroresistance particle counting. The Malvern Mastersizer (MM), an instrument based on laser diffraction, was used without WS. Twelve typical temperate-region soil types were chosen for this study on the basis of their mineral composition, organic matter content, conductivity and magnetic susceptibility. The modalities of PSD frequency curves obtained by different methods were compared. The results showed considerably larger clay-fraction contents determined with the sedimentation techniques, PP and MS, than that obtained by CC or MM.
Statistical correlation and regression models were used to compare the fractions of clay, silt and sand obtained with different methods. Linear correlations were found in almost all cases, except those comparing clay fractions obtained by CC with those obtained by MM, PP and MS methods. These correlations were non-linear. The observed difference was attributed to the presence of aggregates and/or magnetic properties of soil particles. However, more data are needed to verify this hypothesis. The implications of sample pre-treatment with dispersant and hydrogen peroxide (H2O2) were evaluated by the MS and CC techniques and observed by the scanning electron microscope. H2O2 was found to be an efficient disaggregating agent. The use of dispersant resulted in the increased amount of the clay fraction measured by the MS method.