In static models, duality is a convenience. Demand functions cannot generally be determined explicitly from the technology but they are defined implicitly by first order conditions which can serve as the basis for estimation, though perhaps requiring complicated simultaneous equations techniques. Explicit solutions for calculus of variations problems are even rarer and the implicit representation of solutions generally involves a second order nonlinear differential equation (system) and non-trivial boundary conditions. The differential equation system can serve as the basis for estimation only if the generally unrealistic assumption is made that the firm does not revise its plans for several periods and continues along the same optimal path. Thus duality is indispensable for empirical work based on functional forms that are too complicated to be derived directly from the technology as explicit solutions of a problem of intertemporal optimization.