FIG. 8.

The (a) helix content and (b) specific heat from two Monte Carlo simulations are shown. The deviation between the two simulation data for chains of length 10 creates an intrinsic error that prevents us finding a ‘‘good’’ fit when both results are treated as a single solvent type. Ignoring these deviations makes the meaning of goodness not sufficiently restrictive, which allows too many ‘‘good’’ parameter solutions. Instead, these data are treated as two different solvents, where squares and circles represent model-water solvents 1 and 2, respectively.