1 I shall use the word “proposition” wherever Kant uses the word “Satz.”

2 References cite the pagination used in the Academy edition and the standard German abbreviations of Kant's works. G.M.S., p. 417; p. 419.

3 G.M.S., p. 420.

4 G.M.S.,p. 392.

5 In “Analytic Method and Analytic Propositions in Kant's Groundwork,” Dialogue, XV, 4 (1976): 565–82Google Scholar, argued that Kant's assertion that this proposition is synthetic is a mistake and requires a textual emendation. In that paper, however, I did not argue for a formulation of the synthetic proposition which Kant actually sets out to justify. In Ross, W.D., Kant's Ethical Theory: A Commentary on the Grundlegung zur Metaphysik der Sitten (Oxford: Clarendon Press, 1954), p. 72.Google Scholar, Ross argues that the proposition is analytic. In Paton, H.J., The Categorical Imperative: A Study in Kant's Moral Philosophy (New York: Harperand Row, 1967), p. 201Google Scholar, Paton finds Kant's claim that this proposition is analytic puzzling.

6 Patzig, G., Die Logischen Formen Praktischer Sätze in Ethik, Kants, “Kant-Studien, 56, (1965): 237–252.CrossRefGoogle Scholar

7 Beck, L.W., “Apodictic Imperatives,” in Beck, L.W., Studies in the Philosophy of Kant (New York: Bobbs Merrill, 1965): 177–199.Google Scholar

8 G.M.S., p. 414. It is worth pointing out that although Kant initially says (G.M.S., p. 414.) that all imperatives command (gebieten), he later says that only categorical imperatives are commands (Gebote) (G.M.S., p. 416). He also says that whereas all imperatives express necessitation, there is a difference in the necessitation expressed by each kind of imperative (G.M.S., p. 416.), i.e., some express conditioned necessitation whereas others express unconditioned necessitation. I take the explanation of this twofold occurrence of “command” to be the following. When Kant uses “command” as a verb, compare: “All imperatives command…,” he focusses attention on the imperative's expression of the necessitation of an imperfectly rational being to an action. Hence, to say that all imperatives command is to say that each kind of imperative expresses the necessitation of an imperfectly rational being to an action. When, on the other hand, he uses it as a noun, compare: – only the categorical imperative is a command –, he focusses attention on the specific kind of necessitation (namely, unconditioned necessitation) with which the categorical imperative commands. Hence, to say that the categorical imperative is a command is to say that is expresses the unconditioned necessitation of an imperfectly rational being to an action. As a consequence of this twofold use of “command,” Kant can say without redundancy that the categorical imperative commands and is a command, for it (1) expresses the necessitation to an action which is (2) unconditioned. Likewise, he can say without contradiction that a problematic hypothetical imperative commands without being a command, for it expresses the necessitation to an action, which is not unconditioned.

It is also worth noting here that when Kant says: “All imperatives command either hypothetically or categorically” (G.M.S., p. 414.), and, thus, distinguishes them in virtue of the manner in which they command, he is already, although implicitly, drawing attention to another (G.M.S., p. 416.) way of distinguishing them, namely, in virtue of the “Ungleichheit der Nötigung des Willens.” Whereas the imperative that commands hypothetically only commands in virtue of a subjective end, and, thus, can only express the necessitation to an action conditioned by the agent's holding an end; the imperative that commands categorically commands independently of subjective ends and, thus, can continue to express the (unconditioned) necessitation to an action even though the subject may change his subjective ends.

9 Patzig, p. 242.

10 Beck, p. 181. Notice that my example, namely: “Save,” does not have an explicit subject.

11 Patzig, p. 244.

12 This only applies to affirmative categorical judgements as pointed out in Dryer, D.P., Kant's Solution for Verification in Metaphysics (Toronto: University of Toronto Press, 1966), p. 41.Google Scholar

13 Dryer, p. 42., points out condition (3). I have tried to adopt his account of Kant's position on analytic judgements.

14 One of the most severe difficulties in understanding Kant's attempted application of the analytic/synthetic distinction to imperatives is occasioned by a recent, and many think Kantian, way of thinking about value judgements. Many observe that Kant casts moral judgements in the form of imperatives. From this they infer that Kant meant to say that moral judgements are neither true nor false. Consequently, they find great difficulty in applying the analytic/synthetic distinction, which as originally defined necessarily operates on the class of true-false judgements, to those judgements, namely, imperatives, which are not, they urge, in this class.

It is instructive to note that G.E. Moore, writing before the contemporary analysis of moral judgements as imperatives, regarded Kant as a metaphysical moralist, i.e., as one who held moral judgements as asserting an action to have a certain non-natural property, and, thus, to be true or false. Moore regarded Kant as holding this in spite of the fact that Kant held moral judgements to be imperatives; see Principia Ethica (Cambridge: Cambridge University Press, 1962), p. 128ffGoogle Scholar. R.M. Hare, writing during the contemporary analysis of moral judgements as imperatives, on the other hand, rejects Moore's interpretation of Kant and holds that Kant is not a metaphysical moralist. In the end Hare holds that Kant, by calling a moral judgement an imperative, implies that a moral judgement is neither true nor false; see The Language of Morals (Oxford: Oxford University Press, 1964), p. 173, 29.Google Scholar

There are two possible positions here. The first is to presume that by calling an imperative analytic or synthetic, Kant, in spite of contemporary theory, nevertheless regards an imperative as ajudgement which is true or false. As Kant never makes the claim that imperatives are neither true nor false and as he does not point out a need for a new conception of the analytic/synthetic distinction by which it could be applied to judgements which are neither true nor false, the first position will be adopted in this paper. A textual justification for the first position is found in Kant's Opus postumum, V. 21, p. 25. of the Akademie Edition, where he says that the categorical imperative is true:, … des categorischen Imperatives welcher unwidersprechlich wahr ist. … “The second position would consist in an attempt to revise the conception of the analytic/synthetic distinction so that it does not necessarily apply to the class of judgements which are true or false. As I known of no Kantian text which does this, I must regard such an attempt as mere speculation.

15 Paton, p. 124.

16 Nevertheless, the analyticity of the premise has been challenged by Patzig, p. 249. Patzig, however, fails to distinguish, as Kant had done, between “wish” and “want.”

17 Paton, p. 124.

18 Paton, p. 124.

19 Beck, p. 180.

20 Beck, p. 181. “It is not entirely clear in what sense an imperative can be either analytic or synthetic. Imperatives are not judgments with a subject and a predicate, and so do not fall under Kant's explicit division of the types of judgment.”

21 Beck, p. 182. Notice that Beck reads Kant's account in the following way: “If a rational being fullywills B … then the rational being wills A.” (my italics) The word “fully” is, however, correctly omitted by both Paton and Kant. I shall, therefore, omit the word “fully” from my account.

22 A proposition q follows analytically from a proposition p if it either follows formally or follows by an analytic rule of inference.

23 The proposition: “If a rational being is under conditions such that he will act on a categorical imperative, then he will act on a categorical imperative,” is analytic, but not helpful.

24 Beck, p. 182.

25 Beck, p. 182.

26 Beck, p. 182.

27 Beck, L. W., A Commentary on Kant's Critique of Practical Reason (Chicago: The University of Chicago Press, 1960), p. 87. III is the same as (3).Google Scholar

28 G.M.S., p. 417,418. Kant initially distinguishes theoretical cognitions from practical cognitions by saying that whereas theoretical cognitions express what is, practical cognitions express what ought to be (K.r. V., A6₃₃ = B66₁). According to this initial distinction we should expect that whereas theoretical propositions always contain an is, practical propositions (and, thus, imperatives) always contain an ought. Kant seems to confirm this by writing: “All imperatives are expressed by an ought.” (G.M.S., p. 413). There are, however, certain difficulties with this. First, on occasion he expresses imperatives in the imperative mood and, thus, without an “ought.” Hence, not all imperatives are expressed by an “ought.” A second difficulty is the expression of an imperative in the indicative mood without an “ought.” An expression such as: “P is necessitated by reason to do A” can be used to express an imperative, because on Kant's account an imperative is a formula for the expression of the necessitation of an imperfect will by a principle setting forth what it is necessary to do (G.M.S., p, 413). Hence, not all imperatives expressed in the indicative mood are expressed by an “ought.”

In virtue of these points we must revise Kant's: “All imperatives are expressed by an ought” to “All imperatives may be expressed by an ought.” Even though this is so and it forces us to revise Kant's initial distinction between theoretical and practical propositions, it does not touch the heart of the matter. For Kant in using the word “ought” to distinguish practical from theoretical propositions points out that the function of the “ought” is to express the necessitation of an imperfect will by a principle setting forth what it is necessary to do (G.M.S., p. 413). Thus, although the “ought” may be dispensable as a locution used to express the necessitation, it is the presence or absence of the necessitation that is the distinguishing mark between practical and theoretical propositions. It is interesting here to compare K. U., §76, for there Kant points out that it is the subjective constitution of our nature, which is responsible for our being necessitated, that gives rise to the distinction between the practical and the theoretical. Hewrites: “…it is clear that it is only due to the subjective constitution of our practical faculty that the moral law must be represented as a command [and, hence, as necessitating] …, and that reason expresses this necessity [of an action] not as a Sein (Geschehen) but rather as a Sein-Sollen, which would not occur, if reason without sensibility… were regarded as a cause in an intelligible world … where there would be no distinction between Sollen and Tun … and no distinction between a practical and a theoretical law.”

Consequently, I understand Kant's distinction between practical and theoretical propositions as the distinction between those propositions which express necessitation and those which do not, but rather state what is the case. Hence, an indicative proposition not employing the word “ought” such as “P is necessitated to do A” is a practical proposition. This proposition may, of course, also be formulated as “P ought to be A.” Because of this, I shall later hold Kant to the view that practical propositions can be reduced to theoretical propositions if it can be shown that propositions expressing necessitation are inferrable from indicative propositions which do not express necessitation.

29 G.M.S., p. 413.

30 Although I do not wish to make the claim that these three conditions are sufficient for necessitation even in standard cases, I would like to urge that they may be regarded as sufficient for the purposes of the argument. This holds for two reasons: (1) Kant normally shows something by showing that the necessary conditions of it, which are crucial, are met and (2) the subsequent argument will be concerned with a metaphorical application of the concept to a situation in which two persons are not involved.

31 See note 28.

32 G.M.S., p. 400.

33 G.M.S., p. 421; K.p.V., p. 27.

34 G.M.S., p. 420n. Patzig, p. 249, says that there is an obscurity in Kant's claim that the categorical imperative would be analytic if the concept of freedom could be presupposed. On my analysis Kant's statement is not in the least obscure, for it amounts to changing the concept of the addressee to that of an imperfect, yet free, rational being. In the concept of an imperfect, yet free, rational being the concept of the motivation of the will by the universal legislative form of the law is included. Cf. G.M.S., p. 447; K.p.V., p. 29.

35 G.M.S., p. 421.

36 In the third Section Kant has a special argument to circumvent these difficulties. This argument is way beyond the scope of this paper.

37 Paton, p. 244. My italics. Paton writes, p. 201, “Can we justify the proposition that a rational agent as such ought to act in accordance with the principle of autonomy.”

38 Paton, p. 246f.

39 Hare, p. 29.