This paper intends to explain the mechanism of crack propagation and branching in fixed-sided plate, paying attention to its surface roughness and kinetic, dissipative energy. First, considering 2 dimensional problem, the surface roughness, defined using crack path length, and micro propagating velocity, which is along the crack path, can give the upper limit of the feasible kinetic energy. In addition, assuming that the dissipative energy is proportional to surface roughness, we can get 2 dimensional model of branching. Then from the 2 dimensional model, semi 3 dimensional model can be deduced, and comparison of this model with the experimental result of PMMA shows some difference, which is due to existence of discontinuity on actual surface toward the thickness direction. To deal with this problem, correction of semi 3 dimensional model is considered.
