Particle size frequency analysis is used in agronomy, engineering, and geology. The analysis of a sample begins with size classification by sieve opening size S or particle axis length D. The latter includes direct measurements or by sedimentation. The frequency determination of a size class is by size-number, size-mass, or size-volume. Although D and S size units are numerically different and alternative frequency determinations of whole sample geometric mean diameter may differ by a factor of four, there is no rigorous analysis to compare soil and environmental substrate data that are obtained with mixed measurement approaches across disciplines. We began to address this need by using coarse fragments from natural streambeds. We visited 20 Oregon streams and collected 3000 surface and subsurface particles representing a wide range of sizes and shapes. Limiting our research to 4mm<S<256mm (-2φ <S<-8φ), we measured D for each particle along its largest, intermediate and smallest axis lengths, classified it by its sieve opening size S (with size class length ΔS=0.5φ), weighed, and determined its volume. We defined five descriptors using the arithmetic (D_a) or geometric mean (D_g) of the three axes, the intermediate axis length (D_b), the nominal diameter of a sphere (D_n) and the diagonal length of the two smaller axes (D_s). We scale-transformed the D to S units such that S=λD where λ is a constant for each descriptor. We assumed lognormal distribution within each sieved size class (ΔS=0.5φ, or ΔD=0.1φ) and used existing lognormal theory to convert from size-number to size-mass or size volume (and the reverse). We were able to compare whole sample particle size statistics obtained with mixed measurement approaches on equivalent basis. If applied to environmental data, our method will reduce potentially large procedural errors inadvertently introduced in estimated whole sample geometric mean diameter to small statistical errors.