On the use of the Bayliss-McRae dispersion model in NMR solvent effects

The use of the Bayliss-McRae theory on the solvent induced electronic frequency shifts for NMR dispersion shifts is criticized. It is suggested that the NMR shifts should actually be proportional to the square of the Bayliss-McRae function. It is shown that the methane gas-to-liquid shifts in eleven halo-methanes as solvents are indeed proportional to this squared function; ?σ_m(CH₄) = 9.62 (n₂²?1)²/(2n₂²+1)² ppm, where n₂ is the refractive index of the solvent. The relation between this solvent factor and several existing continuum models for NMR medium shifts is discussed.