Geometrical profile (roughness) of joint surfaces influences the behaviour of rock joints under shear loading. With regard to the

dilation, there are two models of direct shear test that may simulate the original loading condition existing in the location from

where the specimens have been sampled. The first model in which the normal load is constant (CNL) and the discontinuity is free

to dilate in shearing, represents typical situations such as movement of a block on a surface slope as a result of its own weight.

The second model in which the dilatancy is prohibited (VNL), simulates the condition of a block confined in a rock mass in an

underground opening. A shear test conducted under restricted normal displacement (dilation) will generally yield considerably

higher shear strength than one conducted under constant normal stress. In this research, both types of tests were conducted on

smooth and rough surfaces of specimens made from rock like material. The results of the VNL and the CNL direct shear tests on

regular teeth-shaped profile discontinuities indicates that at all levels of normal load, the linear Mohr-Coulomb criterion was not

valid for rough surfaces that subscribed to the power law equations. Increasing normal load emphasized the difference between

the results obtained from two methods, although for lower normal loads the results were nearly similar.