Definitive measurements by Van Harlingen et al. in 1980 show that the flux induced by a temperature difference across the two junctions of a Pb-In ring exceeds theoretical expectation by a factor, ϳ105. The theory fails owing to ͑mis͒use of a Boltzmann transport equation to describe the thermal diffusion of quasi-particle excitations, a treatment which violates electron conservation. An electron-conserving transport theory is developed and explains the data.