The vertical velocity field of a natural convection plume arising from a horizontal line heat source was investigated experimentally in detail. The measured vertical velocities were found to be 20–25 percent less than that predicted by analysis. Various causative factors for the discrepancy between analytical and experimental results were investigated, and shown to have a minimal effect. The power law relationship between the plume vertical velocity and energy input to the line source was found to be substantially greater than that predicted by analysis. A normalized similarity function did provide an excellent representation of the lateral diffusion of energy and momentum in the plume, indicating the validity of the boundary layer approximation for the velocity field. It was postulated that further refinement of the stream functions defined by existing analyses is required to accommodate the effects of the upstream flow field that exists in the physical representation of a line source and is omitted from consideration in the development of the analyses.