It is shown that there is an error in Schroeder's calculation of the electronic thermal conductivity due to circulating currents in a two‐phase material. With reasonable values for the properties of the two phases, quantitative agreement between the predictions of the model and the experimental data of Cosgrove, McHugh, and Tiller is obtained for both the electronic thermal conductivity and the Seebeck coefficient.