A transition-state treatment of the relative viscosity of electrolyte solutions is described. The following expression is found for the viscosity B-coefficient: B=(V^°₁–V^°₂)/1000 +V^°₁[(Δµ^{°[graphic omitted]}₂–Δµ^{°[graphic omitted]}₁)/1000 RT]. V^°₁ and V^°₂ are the partial molal volumes of the solvent and solute respectively; Δµ^{°[graphic omitted]}₁ is the activation energy for viscous flow of the solvent, and Δµ^{°[graphic omitted]}₂ the “ionic activation energy” at infinite dilution.

For aqueous solutions, at 25°C, the term (V^°₁–V^°₂)/1000 accounts completely, in the case of ammonium chloride, and partially in the case of potassium, rubidium and caesium chlorides, for the negative sign of the B-coefficient. When solutions in different solvents are compared, the well-known tendency for B to increase with V^°₁ is at least partly explained by the form of the above expression; changes in µ^{°[graphic omitted]}₂ from water to methanol, for example, are less dramatic than changes in B. B-coefficients in the methanol + water system are consistent with a maximum in solvent structure around 20% methanol (w/w) at 25°C; whilst the µ^{°[graphic omitted]}₂-values support this interpretation, it is noted that solvent structure can influence B twice over, in µ^{°[graphic omitted]}₂ and (trivially) in µ^{°[graphic omitted]}₁.