“…Provided that the flow is everywhere transverse to DS in,out ͑E͒, those phase points in the reactant region M A that lie on crossing trajectories 97,98 ͑i.e., that will react, and so are "interesting" in Slater's terminology 10 ͒ can be specified uniquely by coordinates ͑q , p , ͒, where ͑q , p͒ DS in ͑E͒ is a point on DS in ͑E͒, the incoming half of the DS, specified by 2͑n −1͒ coordinates ͑q , p͒, and is a time variable. ͑Divid-ing surfaces constructed by the NF algorithm are guaranteed to be transverse to the vector field, except at the NHIM, where the vector field is tangent.…”