Streamline advection has proven an eeective method for visualizing vector ow eld data. Traditional streamlines do not, however, provide for investigating the coarser-grained features of complex datasets, such as the white matter tracts in the brain or the thermal conveyor belts in the ocean. In this paper, we introduce a cohesive advection primitive, called a stream bundle. Whereas traditional streamlines describe the advection patterns of single, innnitesimal micro-particles, stream bundles indicate advection paths for larger macro-particles. Implementationally, stream bundles are composed of a collection of individual streamlines (here termed bers), each of which only advects a short distance before being terminated and re-seeded in a new location. The individual bers combine to dictate the instantaneous distribution of the bundle, and it is this collective distribution which is used in determining where bers are re-seeded. By carefully controlling the termination and re-seeding policies of the bers, we can prevent the bundle from becoming frayed in divergent regions. By maintaining a cohesive form, the bundles can indicate the coarse structure of complex vector elds. In this paper, we use stream bundles to investigate the oceanic currents.