A multicomponent mixture may be viewed conceptually as a hypothetical collection of fluid clusters. In this context a mixture model is defined by prescriptions for (a) estimating fluid cluster properties and (b) combining them to yield an overall mixture property. A particularly flexible form is obtained using composition-weighted power means with the weighting based on global mole fractions. It predicts multicomponent properties from knowledge of pure component and binary mixture data. The classic quadratic Scheffe rational polynomials and the Wilson models are special cases. The model also generates a cubic Scheffe form for which ternary and higher coefficients can be expressed in terms of binary interactions. The binary-correlative and multicomponent-predictive capabilities of the model were evaluated for isothermal liquid viscosity data. A revised empirical mixing rule for liquid viscosity is proposed for systems where the constituent binaries show either convex or concave composition dependence.