In section 6 of their paper, Effects of Government Expenditure on the Term Structure of Interest Rates (Journal of Money, Credit, and Banking 16 [February 1984], 16-33), Turnovsky and Miller investigate the conditions under which the dynamical system generated by a bond-financed fiscal expansion will possess a stable manifold with a dimension equal to the number of backward-looking variables. Since in the reduced form of their model there is one forward-looking variable, P, the price of bonds, and one backward-looking variable, B, the quantity of bonds, and since they consider a system linearized about its steady state, this question reduces to investigating whether the determinant of the system's state matrix is negative: if it is, then the system has a saddlepoint. The model that Turnovsky and Miller use is a variant of the familiar BlinderSolow model, and it is well known that for a bond-financed fiscal expansion to be stable in that model the following condition must hold: an increase in the supply of bonds will generate an increase in income large enough for the increased tax receipts to more than finance the extra coupon payments. I argue in this note that the Blinder-Solow condition, properly translated from the original model (with one rate of interest) to the Turnovsky-Miller variant (with two rates, connected by rational expectations), is more stringent than Turnovsky and Miller imply in their paper and that it is a necessary and sufficient condition for the linear dynamical system in P and B to possess a saddlepoint. This is in contrast to the conclusions of Turnovsky and Miller, who state that it is necessary, but not sufficient. In the Turnovsky-Miller model, instantaneous equilibrium is determined by the solutions to the lS-LM equations; these solutions are given by