In the second part of their paper, Weaver and Viskanta (1991b) aimed to demonstrate the effects of the interdiffusion and Dufour energy fluxes and the Soret vapour flux. As in the first part of the paper (1991a), the vehicle was a series of numerical solutions for plane vertical square cavities.
As explained in §3.3.12, the treatment of interdiffusion was fundamentally flawed. Further results (neglecting the Soret and Dufour effects; their inclusion would violate the hypotheses of Theorem 1) showing strong relative extrema of temperature were presented. This phenomena was erroneously ascribed to interdiffusion (its occurrence in the solutions of 1991a was supposedly due to the inclusion of interdiffusion at the boundaries).
The treatment of the Dufour effect was marred by the fact that the expression presented for the diffusion-thermo flux did not have the same dimensions as the other energy fluxes (conduction, interdiffusion and bulk advection).
Lacking experimental data and doubting the accuracy of theoretical predictions, Weaver and Viskanta assigned arbitrary values to the thermal diffusion factor. This detracts from the usefulness of the study. Bizarrely, in some runs the thermal diffusion factor was assigned different values in the thermal diffusion and diffusion-thermo fluxes; this contradicts Onsager's reciprocal relation (Bird et al. 1960, p. 564).
The conclusions drawn by Weaver and Viskanta (1991a, b) are vague, qualitative and often erroneous.