In many earthquakes, aftershock activity tends to cluster near the endzone of the fault. This clustering is expected from elastic theory, i. e., two dimensional crack causes high stress region near two endzones of the fault. But, in many shallow strike-slip earthquakes, we can find only one of them. In this paper, we will solve why the clustering is activated near either of the endzones. We study aftershock distributions and source processes of inland long, shallow faults in which we can easily distinguish the clustering near the endzones. It is shown that in bilateral case we can see two clusters at two endpoints of rupture propagation, but in unilateral case one cluster at an endpoint of rupture. The dynamic crack model shows that accelaration, particle velocity and dynamic stress change near the stopping point of rupture are larger than those near the starting point. This may cause much activation of aftershocks near the stopping point of rupture than near the starting point, i. e., hypocenter of main shock.
