1 References to Kant's, works in the pagination of the Royal Prussian Academy edition of Kant's Gesammelte Schriften (Berlin: 1902-1966)Google Scholar, hereafter Ak, are given in the text according to the following abbreviations: CPR: Critique of Practical Reason (Ak, Vol. V); MM: Metaphysics of Morals (Ak, Vol. VI). References to the Groundwork (Ak, Vol. IV) are enclosed in brackets without abbreviation.

2 Kant's second set of derivations relies upon the third or ‘End-in-Itself’ formulation of the categorical imperative (492), and it has been argued that we must look to it and its teleological notions if we are to understand the first set of derivations (cf. H.J. Paton, The Categorical imperative: A Study in Kant's Moral Philosophy [New York 1965], 148-52). But Kant himself certainty thought that specific duties could be derived from the universal law of nature formulation without recourse to the third formulation (423); furthermore, he believed the two formulations to be equivalent (436), so any logical consequence of one should be a logical consequence of the other. In fact, although Kant's notions of an end-in-itself and harmony (430) are important, they seem too elusive to allow us to determine whether the two formulations are really equivalent or not; by comparison, the notion of consistency which figures so prominently in the first set of derivations is reasonably straightforward.

3 Proponents of this view include Ewing, A.C. ‘The Paradoxes in Kant's Ethics, Philosophy, 13 (1938) 40·56CrossRefGoogle Scholar, p. 49 (though Ewing believes Kant's use of empirical premises to be inadvertent); Paton, 146-51; Harrison, Jonathan ‘Kant's Examples of the First Formulation of the Categorical Imperative,’ Philosophical Quarterly, 7 (1957) 50–62CrossRefGoogle Scholar, passim; Hill, Thomas Jr. ‘The Hypothetical Imperative,’ Philosophical Review, 80 (1973) 429–50CrossRefGoogle Scholar, esp. pp. 442-4; Onora Nell, Acting on Principle: An Essay in Kantian Ethics (New York 1975), esp. pp. 66-81; Potter, Nelson Jr. ‘How to Apply the Categorical Imperative,’ Philosophia, 5 (1975) 395–416CrossRefGoogle Scholar, esp. p. 397; and Buchanan, Allen ‘Categorical Imperatives and Moral Principles,’ Philosophical Studies, 31 (1977) 249–60CrossRefGoogle Scholar, passim. Somewhat related views may be found in Nicholas Rescher's ‘Kant and the Special Constitution of Man's Mind,’ International Kant Congress (1974), 318-28.

4 Views of this sort have been defended by Marcus, Singer in ‘The Categorical Imperative,’ Philosophical Review, 63 (1954) 577–91Google Scholar, and Buchanan.

5 The necessity of the background assumptions is all that is needed if duties are to be unconditional. That the necessity take the form of analyticity is required only for the claim that the universalization of a maxim contravening a perfect duty is self-contradictory.

6 However caution is suggested by Kant's apparent vacillation between viewing self-love and happiness as two sides of the same coin and his allowing that at least some people may derive happiness from various actions without any motive of self-interest (e.g., 398). The casuistical questions on suicide in the Metaphysics of Morals (423-4) also give room for pause, though the argument against suicide in the later work is rather different from the one Just examined, and I am inclined to think that the two yield conclusions of differing generality. It may be noted that insofar as Kantian duties have the sort of generality suggested above, the vexed question about the relevant descriptions of actions in agent's maxims becomes a less serious problem.

7 It seems likely that Kant would wish to attach the qualification that such a promise will be made only when an agent believes doing so has some chance of relieving his need. For simplicity I shall omit this provision, though we will see that it may be important (f.n. 16). For brevity, I shall call a promise made with no intention of keeping it a false promise.

8 That the contradiction is between the law and the agent's willing to make a false promise has been urged by Ewing, 44-5 and Nell, 78.

9 Kant's wording suggests the cumbrous talk of sorts of promises. These assumptions, or near variants of them, have often been said to be contingent, cf. Ewing, 49, Paton, 152, and Harrison, 55.

10 This is not simply an artifact of the present interpretation. Kant's own argument can be adapted to show that if it were a universal law that we make false promises to save innocent lives such promises would be greeted as vain pretenses, that the law would therefore be self-contradictory, and, hence, that we should never make them. That Kant might have agreed is perhaps suggested by the hard line he takes in his essay ‘An Alleged Right to Lie from Beneficent Motives’ (Ak, VIII, 425-30). But his arguments there are neither impressive nor clearly related to those in the Groundwork; moreover, commentators have provided good reasons for doubting that this essay reflects Kant's considered view.

11 The difference between actual willing and rational willing can be more clearly expressed in terms of Kant's later distinction between two aspects of the will, namely Willkür, or the power of choice, and Wille, or the rational power of selflegislation (e.g., MM, 13-14). This distinction is not explicitly drawn in the Groundwork, but it is clearly foreshadowed in Kant's distinction between the pure will or practical reason, on the one hand, and the will as affected by inclinations and desires, on the other (454; cf. 412, 424, 458).

12 This provision can plausibly be built into all maxims tested, thus removing the appearance that a duty applies to a rational being only if he has certain abilities and — since it is an empirical question what abilities he has — that empirical information is needed in determining his duties. Rather, his duty is to do whatever is within his power to help others, to develop his talents, or the like.

13 An illuminating way of making this point is Robert Paul Wolff's proposal (in The Autonomy of Reason: A Commentary on Kant's Groundwork of the Metaphysics of Morals [New York 1973], 170) that it turns on an application of the dictum, regarded by Kant as analytic, that insofar as one is rational, when he wills an end he also wills the means within his power necessary to attaining it (417). Since the man in our thought experiment does, by hypothesis, have the means within his power to make it possible that he receive aid by legislating a law of nature that others be beneficent, a failure to do so would be irrational.

14 This has been noted by Ewing, 49 and Paton, 152. We will ask whether similar objections apply to our most recent interpretation of the third example once we have examined this claim.

15 A fact Kant perhaps recognizes seven pages later when he tells us that ‘many a man would readily consent that others should not help him, if only he should be excused of showing beneficence to them’ (430n). Similarly, in regard to our recent reinterpretation of Kant's third example, it would be possible for a being to set or retain only ends that could be achieved without cultivated talents (cf. 423). A different but equally serious problem arises in the case of false promising, for though all dependent rational beings have needs, it would seem to be an empirical question whether they have needs they believe they could meet only by making false promises (cf. f.n. 8). If so, the emendation of the fourth derivation suggested below will need to be extended to the derivation of the second example as well as to the third. Ewing (46) hints at and Wolff (170-1) notes the possibility of never setting or retaining ends that could not be attained without help, though neither link this to the present points about will and desire, talents, or false promises.

16 I am indebted to Richard Eggerman, Barbara Levenbook, Terrence McConnell, W.H. Walsh, and the participants at a Colloquium at the University of Kansas for helpful comments on an earlier version of this paper.