Two alternative ways to treat the entropy of solution in the molecular thermodynamics of liquid-phase solutions have proved useful under various circumstances. The first is the ideal solution theory in which the entropic effects are the same as for mixing ideal gases. The second is the theory of Flory and Huggins which is most appropriate for nondilute solutions of chain molecules in small solvents. The two theories are compared for their ability to describe solutions of alkanes, both straight-chain and branched, in aqueous solution. The alkanes provide an especially appropriate testing field because electrostatic effects are minimal, and it is reasonable to assume that the solvation free energy consists almost entirely of entropic contributions and first-hydration-shell effects. The tests show that the experimental data are better correlated by the ideal solution theory than by adding an explicit volume-dependent contribution from Flory-Huggins theory as suggested by Sharp, Nicholls, Friedman, and Honig. We disagree with a recent recommendation that one should use Flory-Huggins theory to change the definition of the experimental free energy of transfer of a solute from the gas phase into solution and also with the suggestion that there is a general volume contribution to the free energy of solvation that is well modeled by the Flory-Huggins term.