These two formulations are different from each other except that a constant H happens to be a Samuelson optimal value. In other words, Strotz's optimization is more constrained than Samuelson's, and Strotz's optimization cannot achieve Pareto optima except in the special case just mentioned.2 Does this difference between the two approaches invalidate Strotz's propositions in section II and section III in his article? If we confine our problem to Pareto optima, Strotz's propositions happen to be meaningful, that is, when H happens to be Samuelson's optimal value. And even in this case the relation between the optimal value of H and Pareto optima is ambiguous, and we must explicitly treat productive services of fixed income assets in the model. But it can be expected that such an approach would introduce complications into the model. Although Strotz's formulation has these difficulties, I believe, Samuelson's comment on Strotz's article, in the Appendix to Aspects of Public Expenditure Theories in this REVIEW, XL (November I958) is valid, that is, changing public goods does materially affect the distribution of income and all decisions have to be made simultaneously.