In order to express the particle behaviour in a pipe flow, the vertical-and longitudinal-probability density functions of particles should be adopted. The method using those functions to express the particle behaviour in a pipe flow was already adopted by Binnie, Barnard and Sharp. They supposed that the particle movements were Markov's process and that suspension and dispersion of particles were caused by turbulent diffusion of a pipe flow. The authors have analysed and tested the particle movements on the basis of assumption that suspension and dispersion of particles have been caused mainly by lift of shear-flow and particle saltation at the pipe wall, especially with coarse particles, rather than by a turbulent diffusion of the pipe flow. The test apparatus has a 22 m closed conduit pipe of 52 mm diameter with a transparent test section. The particles used are 6 mm diameter nylon spheres.