Sharp et al. have raised the question : If a solute and its solvent have different molecular sizes, how should solvation interaction energies be extracted from transfer experiments? Is the chemical potential more appropriately given by "classical" solution theory or by Flory-Huggins theory? We study a general statistical mechanical treatment of solvation, a generalized Flory-Huggins theory, and related treatments by Hildebrand and Sharp et al. to determine the physical basis for size-dependent terms in the chemical potential, and their limitations and applicabilities. We find that the extra entropy in the Flory-Huggins theory does not arise from the disparity of sized of solute and solvent, from free volume, or from artifacts of approximations. Rather, when solutes and solvents have sufficient complexity that they can "interfere" with each other in solution, there is an "entropy of coupling" of translational freedom to excluded volume or internal or rotational degrees of freedom. The Flory-Huggins theory approximates this coupling entropy for polymeric solutes or solvents, but not for other systems. Proper extraction of contact free energies requires proper subtraction of the coupling term. This study rationalizes several experimental and simulation studies, and indicates that coupling entropies that depend on molecular size and shape are often needed to treat complex solvation processes.