1. men the analysis describes a motion starting f r m rest, there are no sources of perturbation at infinity. In one-dimensional problems, a shockless perturbation front is followd by a simple wave w i n g outwards; it is, thus, very easy to calculate boundary values on any artificial boundary at a finite distance, when the boundary is C K O S S y a simple wave. An extension of the idea to multi-dimensional problems may be pssible, but I have no personal experience on the subject ard I have no notion of similar attempts by others so far. 12. m i l d i q a computational grid