Particle Size and Texture Analysis

Sieve analysis is the standard method for particles above 63 micrometres (the boundary between sand and silt). The sample is pre-treated to remove organic matter and to disperse aggregates, then oven-dried to a constant mass. It passes through a stack of ISO or ASTM sieves arranged from coarsest to finest, usually at 2 mm, 600 micrometres, 250 micrometres, 125 micrometres, and 63 micrometres, with the exact series chosen by the laboratory protocol.

After mechanical shaking for a standard period, each sieve is weighed and the retained mass expressed as a percentage of the total. The cumulative percentage passing each sieve is plotted against the sieve aperture on a semi-logarithmic scale to produce the grading curve. The D50 (median grain size), the D10 and D90 percentiles, and the coefficient of uniformity are standard statistical descriptors read from this curve.

Silt (2-63 micrometres) and clay (below 2 micrometres) are too fine to retain on a sieve. The classical solution uses sedimentation. When a deflocculated soil suspension is mixed in a cylinder, particles settle at a rate governed by Stokes' Law: settling velocity is proportional to the square of the particle diameter and the density difference between particle and water. Large particles settle fast; clay platelets take hours or days.

In the hydrometer method, a calibrated hydrometer measures the density of the suspension at timed intervals. As particles settle out, the density falls. From the density readings at 40 seconds, 2 hours, and longer intervals, the proportions of sand, silt, and clay are calculated using the Bouyoucos or ASTM formulae. The pipette method withdraws measured volumes from a fixed depth at timed intervals, dries them, and weighs the residue. Both methods depend on complete dispersion of the sample using sodium hexametaphosphate, and both require temperature control because the water viscosity term in Stokes' Law changes with temperature.

Laser-diffraction instruments work on a different principle entirely. A soil sample is dispersed in water (or, for very fine particles, a non-aqueous solvent) and pumped past a laser beam. Particles scatter and diffract the light at angles that depend on their size, with small particles producing wide-angle diffraction and large particles producing narrow-angle diffraction. An array of detectors captures the angular intensity pattern, and software inverts it using Mie or Fraunhofer scattering theory to produce a volume-weighted PSD.

The Malvern Mastersizer series (various generations since the 1990s) has become the reference instrument in forensic geology laboratories, used extensively in the work of Kenneth Pye and colleagues at the Kenneth Pye Associates and Royal Holloway University of London. Key advantages for forensic work: the measurement takes under a minute, the sample requirement is milligrams, the output is a full high-resolution curve covering clay to coarse sand in one run, and the instrument log is a documentary record.

Once the percentages of sand, silt, and clay are measured, they can be plotted on a ternary textural triangle. The USDA system, widely used in North America and adopted in many forensic references, divides the triangle into twelve named classes: clay, silty clay, sandy clay, clay loam, silty clay loam, sandy clay loam, loam, silt loam, silt, sandy loam, loamy sand, and sand. The ISSS (International Soil Science Society) system uses different size boundaries: silt extends from 0.002 mm to 0.02 mm (not 0.05 mm), so the sand fraction is larger in ISSS measurements.

For forensic casework, the textural class gives quick communication, but it is rarely the comparison unit. A clay loam covers a wide swath of composition space and includes many unrelated soils. The comparison that matters is between the full PSD curves, or between statistical parameters extracted from them: D10, D50, D90, the sorting coefficient, and the skewness. Two samples that plot at the same point in the textural triangle are genuinely similar; two that merely share a class name may not be.

Pedologists doing field mapping have long assessed texture by feel. A small amount of soil is moistened in the palm of the hand until it is plastic but not sticky. The soil is then pressed between thumb and forefinger and squeezed forward to form a ribbon. A pure clay produces a long, smooth, continuous ribbon over 25 mm that may reach 75 mm or more. A sandy loam forms only a short, broken ribbon or none at all. A loam falls in between.

In forensic contexts, the ribbon test has two legitimate uses. At the scene, it gives an examiner an immediate rough sense of what they are dealing with, so that sample containers and preservation methods can be chosen appropriately. In the preliminary examination stage, it screens samples to decide whether detailed analysis is warranted. It is not a method for formal comparison because ribbon length and feel vary with operator experience and the exact moisture content at the time of testing.

The forensic value of particle-size analysis rests on how variable PSD is across real terrain. Soil-forming processes, geology, transport history, and land management all leave signatures in the distribution. A fluvial sand deposited by a river has a characteristically sorted, unimodal distribution peaking around medium sand. A glacial till is poorly sorted, spanning a wide range. A loess is fine-grained, predominantly silt-sized, and tightly clustered around 20-40 micrometres. These differences are large enough to be diagnostic of origin even before any chemistry is done.

In case practice, Pye and Blott have demonstrated that laser-diffraction PSD curves, combined with statistical comparison using the Kolmogorov-Smirnov test or discriminant analysis, can separate soils from adjacent fields on the same farm. The discrimination depends on having a well-characterised reference dataset. When a questioned sample falls close to the reference cluster but is also close to other soil types in the area, the PSD result needs support from mineralogy and chemistry to carry evidential weight.