A limit point p of a discrete group of Möbius transformations acting on S n is called a concentration point if for any sufficiently small connected open neighborhood U of p, the set of translates of U contains a local basis for the topology of S n at p. For the case of Fuchsian groups (n = 1), every concentration point is a conical limit point, but even for finitely generated groups not every conical limit point is a concentration point. A slightly weaker concentration condition is given which is satisfied if and only if p is a conical limit point, for finitely generated Fuchsian groups. In the infinitely generated case, it implies that p is a conical limit point, but not all conical limit points satisfy it. Examples are given that clarify the relations between various concentration conditions.Definition: An open set U in S m−1 can be concentrated at p if for every neighborhood V of p, there exists an element γ ∈ Γ such that p ∈ γ(U) and γ(U) ⊆ V . If in addition the element γ can always be selected so that p ∈ γ(V ), then one says that U can be concentrated with control.Note that U can be concentrated at p if and only if the set of translates of U contains a local basis for the topology of S m−1 at p. Also, one can easily check from the definition that (1) there exists a neighborhood of p which can be concentrated with control if and only if there is a connected neighborhood which can be concentrated with control (take the connected component of U that contains p, and require that γ −1 (p) ∈ U ∩ V ), and (2) if a neighborhood of p can be concentrated with control, then every smaller neighborhood can be concentrated with control.Definition: The limit point p is called a controlled concentration point for §1. Introduction 3 Γ if it has a neighborhood which can be concentrated with control at p.Concentration with control is studied in [1]. Analogously to conical limit points, p is a controlled concentration point if and only if there exist a point r = p in S m−1 and a sequence γ n of distinct elements of Γ so that γ n (p) → p and γ n (x) → r for all x ∈ S m−1 − {p}. In particular, every controlled concentration point is a conical limit point. However, examples are given in [1] of conical limit points of 2-generator Schottky groups which are not controlled concentration points (see also proposition 4.1 below). For groups of divergence type, controlled concentration points have full Patterson-Sullivan measure in the limit set. There is a direct connection between controlled concentration points and the dynamics of geodesics in the hyperbolic manifold B m /Γ. Call a geodesic ray in B m /Γ recurrent if it is the image of a geodesic ray in B m that ends at a controlled concentration point. In an appropriate metric, the space of recurrent geodesic rays in B m /Γ is a metric completion of the space of closed geodesics in B m /Γ (where both spaces are topologized as subspaces of the unit tangent bundle of B m /Γ).We turn now to weaker concentration properties. It is not difficult to show (see [6]) that every limit point p ...