We have investigated the influence of the steric properties of conducting
particles in a nonconducting host matrix on the conductivity threshold of
the material, i.e., the minimum volume fraction of conducting phase for the
whole sample to become conducting. A statistical, numerical method is used
in which the particles are randomly put, one by one, into the nonconducting
host and the conducting path is searched. The particles are allowed to
penetrate each other to some extent. Three different types of particle
shapes are considered: spherical, cylindrical with rounded ends and
asymmetric cuboids with rounded surfaces. We have found that in addition to
the anisotropy in the particles' dimensions, the angular distribution of the
particles' long axes plays a dominant role in the calculations of the
conductivity percolation threshold.