^{page 434 note 1} Or “phenomenal” objects. Though Kant is thinking mainly of the physical world, “physical” must not be taken as excluding “psychical” in what follows, so far as by psychical is meant what is known in introspection. Actually the question of the application of the categories by Kant to the phenomenal self is a complicated one; see Ewing, Kant's Treatment of Causality, Ch. VI.

^{page 435 note 1} Third Dialogue, p. 287 (Everyman).

^{page 435 note 2} Treatise, I, iv, 2.

^{page 435 note 3} Ibid., p. 207, Selby—Bigge.

^{page 435 note 4} This also seems to be the implication of Berkeley's nominalism.

^{page 435 note 5} If the general law of causality is really derived from experience, it should not extend further than experience has shown it to extend. But Hume treats it as if it had universal validity, e.g. in his remarks about miracles and contrary causes. Certainly Hume says that anything may be the cause of anything; but unless we could guarantee some order in the appearance of sense-data, could we form any causal laws? Kant's theory attempts to show that we can expect a degree of stability in the physical world without committing ourselves to any metaphysical remarks about the source of sense-data.

^{page 437 note 1} For the sake of simplicity I have followed the apparent indications of the first half of the first edition deduction here. This may result in a distortion, since the pure synthesis may not be of space and time themselves but of objects in space and time. But this is not the place to discuss details of Kantian interpretation, and dogmatism is unavoidable.

^{page 439 note 1} And stands in relation to other objects, forming with them an objective world. It must not be thought that “first” and “second” here refer to temporal order.

^{page 439 note 2} And in any case Kant says (B 145) that the categories would be of no use to an intelligence possessed of intellectual intuition, presumably because such an intelligence would not need to synthetize manifold given from without (perhaps, would not need synthesis at all).

^{page 440 note 1} Yet factual propositions depend on prescriptive propositions, since without the latter there could be no facts.

^{page 440 note 2} In Kant mathematical propositions are clearly quasi-factual, since they describe the essential nature of space and time as well as prescribing laws which everything which falls within space and time must obey.

^{page 441 note 1} There are also divisions within the class of specially prescriptive propositions. Kant, e.g. regards arithmetic as more generally prescriptive than geometry.

^{page 441 note 2} This argument as it stands is unduly dogmatic (i) because there seem to be some factual propositions which come very near to being a priori, e.g. “there are categories,” “sense and thought co-operate in knowledge” (and, in fact, philosophical propositions generally); (ii) because it rests on a dogmatic denial of the very thing metaphysics claims—that the senses are not the only source of knowledge. Metaphysics does allege a non-empirical source for some of its propositions (e.g. pure apperception for rational psychology), and this allegation ought to be investigated. But all this does not alter what has been said about the principles.

^{page 442 note 1} Using “object” in a non-technical sense (as Kant sometimes does, often with confusing results; cf. e.g. B 122 = A 89).

^{page 442 note 2} Assuming that every discursive intelligence has a sensibility. Every such intelligence has a “receptivity,” but it is not clear whether this is the same thing as a sensibility. Kant sometimes appears to mean “sensibility like the human” by “sensibility” (cf. e.g. B 342–3 = A 286–7). The whole question of the possibility of other types of intelligence than ours is a complicated one, and Kant perhaps contradicts himself (cf. B 155 with B 139 and B 283 = A 230, and again with B 43 = A 27).

^{page 443 note 1} B 130.

^{page 444 note 1} The ideas of reason are further removed from sense, but it is ideas that we seek to determine in thinking objects in general.

^{page 445 note 1} B 303 = A 246–7.

^{page 445 note 2} It accounts for Ontology in the Wolffian division.

^{page 446 note 1} B 705 = A 677.

^{page 447 note 1} Except in formal logic, and that is not knowledge in the strict sense.