D Starting with knowledge of the internal energy of a polyatomic system as a multidimensional function of the coordinates of the component atoms-the effective potential surface, one can explore the topography of this surface to find its stationary points and topology. Clusters are particularly effective vehicles for developing the methods of doing this, although it is essentially as easy to study small molecules as it is clusters, if the corresponding surfaces are already known. If the system is small enough, all the minima, the relevant saddles, and the corresponding reaction paths can be found; if the system is larger, then only statistical sampling methods can be applied. Such explorations can be used as efficient ways to test the physical plausibility of potential surfaces, e.g., to determine whether a surface developed to describe spectra is valid enough globally to be used for scattering studies. With the readily manageable potentials such as that composed of the sum of pairwise Morse interactions, it is now straightforward to explore how the form of the basic pair interactions affects the multidimensional topography of the surface. For systems of many particles, it is beginning to seem feasible to infer from topographical properties of the surface the extent to which a system undergoing cooling is either glass-forming or "focusing," in the sense of going to a single structure or a small set of related structures. 0 1996 John Wiley & Sons, Inc. like to think that the topic of multidimensional potential surfaces is difficult and deep enough to appeal to his excellent but demanding taste. Here, we have an old subject, but still it is one in which we are just now trying to conjure what the useful questions should be. We have enormous power to do computations and to generate great volumes of numbers, but what useful information do all these