Abstract
Two conceptual paradigms have been used in the past to interpret the observed strength and structure of eddy heat fluxes in the atmosphere. The relative merits of these two approaches are examined using a two-level primitive equations model on a sphere.