<i1> [B1] Introduction
1. On the difference between pure and empirical knowledge
There is no doubt whatever that all our knowledge begins with experience, since our faculty of knowledge could not be awakened into activity unless objects affected our senses. These objects partly produce representations by themselves, and partly stimulate the activity of our understanding into comparing, combining, or separating these representations, thus converting the raw material of sensory impressions into a knowledge of objects which is called ‘experience’. So, in the order of time, we have no knowledge before experience, and all knowledge begins with experience.
<i2> But although all our knowledge begins with experience, it does not follow that it all originates from experience. For it could well be the case that even our experiential knowledge is a combination of what we get from impressions, and what our own faculty of knowledge supplies from its own resources, merely on the occasion of sensory impressions. However, we cannot distinguish this additional element from the [B2] basic material until lengthy practice has made us sensitive to it, and skilled at separating it out.
<i3> So there is at least a question which needs closer investigation, and which is not to be rejected out of hand. This is the question whether there is such knowledge that is independent of experience, and independent even of all impressions of the senses. Such knowledge is called apriori, as contrasted with empirical knowledge, which has its sources aposteriori, namely in experience.
<i4> However, the expression ‘apriori’ has not yet been defined precisely enough to bring out the full meaning of the question at hand. For we customarily say of much knowledge derived from experiential sources that we have it or are capable of having it apriori, because we do not derive it directly from experience, but from a general rule — even though we have borrowed the rule itself from experience. For example, we might say of someone who was digging under the foundations of their house, that they could know apriori that it would collapse; in other words, they need not wait for the experience of its actually collapsing. However, they could not know this completely apriori, since they would already have to have known through experience that bodies are heavy, and hence that they fall when their support is removed.
<i5> So in what follows, I shall not mean by ‘apriori knowledge’ knowledge which is independent of this or that [B3] experience, but knowledge which is absolutely independent of all experience. Its opposite is empirical knowledge, or knowledge which is possible only aposteriori, i.e. through experience.
<i6> When apriori knowledge contains no empirical element whatever, it is called pure. So, for example, the proposition ‘Every alteration has its cause’ is an apriori proposition, but it is not pure, because the concept of alteration is one which can be derived only from experience.
<i7> 2. We have some apriori knowledge, and even the ordinary understanding is never without it
What we need here is a reliable criterion for distinguishing between pure and empirical knowledge. Experience certainly teaches us that something has such-and-such a nature, but not that it could not be otherwise. So first, if we have a proposition which cannot be thought without being thought as necessary, then it is an apriori judgment; and if, in addition, all the propositions from which it is derived are themselves valid as necessary propositions, then it is absolutely apriori. Secondly, experience never provides its judgments with genuine or strict universality, but only with a hypothetical and relative universality depending on induction. All we can justifiably say is that, as far as we have observed up until now, [B4] there has been no exception to this or that rule. So if a judgment is thought with strict universality (i.e. so that no exception at all is allowed as possible), then it is not derived from experience, but is valid absolutely apriori.
<i8> Empirical universality is merely the unjustified raising of the status of a claim which is valid in most cases, to one which is valid in all cases — for example, in the proposition ‘All bodies are heavy.’ When, on the other hand, a judgment essentially involves strict universality, this indicates a special source of knowledge, namely a faculty for apriori knowledge. Consequently, necessity and strict universality are reliable criteria for apriori knowledge, and are inseparably connected with each other. In practice, however, it is sometimes easier to show that judgments are limited to experience than that they are contingent, and it is often more convincing to show that a judgment has unlimited universality than that it is necessary. So it is advisable to use both the above criteria separately, given that each is infallible by itself.
<i9> Now, it is easy to show that human knowledge does include such judgments, which are both necessary and universal in the strongest sense, and hence are pure apriori judgments. If you want an example from the sciences, you need only consider the propositions of mathematics. If you want an example from the most everyday use of the understanding, [B5] the proposition that every alteration must have a cause will serve the purpose. Indeed, in this last proposition, the concept of a cause so obviously contains the concept of the necessity of its connection with an effect, and the strict universality of the causal law, that the concept would be completely destroyed if you tried to follow Hume’s derivation of it. Hume attempted to derive it from a repeated association of what happens with what precedes it, and a resulting habit of connecting the representations of them — with the consequence that the necessity is merely subjective.
<i10> In fact we do not need examples like these in order to prove that our knowledge actually includes pure apriori propositions, since it can be proved by their indispensability for the very possibility of experience, and hence proved apriori. Where would experience get its certainty from, if all the rules by which it changes from one state to the next were as empirical as experience itself, and hence contingent? Contingent rules could hardly count as primary axioms. However, at this stage it is enough to have established the pure use of our faculty of knowledge as a fact, together with the criteria for it.
<i11> It is not only judgments that reveal themselves as having an apriori origin, but also even some concepts. If you abstract from your empirical concept of a body, one by one, its colour, its hardness or softness, its weight, and even its impenetrability, there still remains the space which it occupied (even though the body itself has now entirely disappeared), and [B6] you could not leave that out of your concept of the body. Similarly, if you leave out of your empirical concept of any object, whether material or immaterial, all the properties which you have learned about through experience, you still cannot deprive it of the property through which it is thought as a substance, or as inhering in a substance — even though this concept is more determinate than that of an object in general. So, convinced by the necessity with which this concept forces itself on you, you must admit that it is located apriori in your faculty of knowledge.
<i12> 3. Philosophy needs a science which will determine apriori the possibility, the principles, and the scope of all apriori knowledge
What is even more important than everything I have said so far is that some knowledge seems even to transcend the domain of all possible experiences, and to broaden the reach of our judgments beyond all limits of possible experience, through concepts for which no corresponding object at all can be given in experience.
<i13> This sort of knowledge, which transcends the world of the senses, and where experience can neither guide us nor correct us, is precisely the area investigated by our reason. We [B7] consider these investigations to be much more important and much more exalted in their aims than anything which the understanding can learn within the domain of appearances. We would rather run every risk of error, than give up such important investigations because of some grounds for caution, or because we might consider them of little value or a matter of indifference.
<i14> These unavoidable tasks for pure reason itself are God, freedom, and immortality. The science which, after all its preliminary work, is essentially directed towards solving just these problems, is called metaphysics. In the earliest stages of its development, the procedure of metaphysics is dogmatic — that is, it undertakes the execution of its task with confidence, and without first assessing the capacity or incapacity of reason for such a major undertaking.
<i15> Now it seems perfectly natural that, as soon as you have left the firm ground of experience, you would not immediately erect a building out of knowledge you possess without knowing where the knowledge came from; nor would you trust in axioms you did not know the source of. Rather, you would first assure yourself of the foundations of the building through careful investigations. In other words, you would start by raising the question of how the understanding can come by all this apriori knowledge, and also the question of what scope, validity, and worth it might have. In fact, nothing is more natural, if you take ‘natural’ in the sense of what ought to happen in the most appropriate and rational way.
<i16> [B8] But if by ‘natural’ you mean what usually happens, then the situation is reversed, and there is nothing more natural or understandable than that this investigation should have been neglected for so long. One branch of this kind of knowledge, namely the mathematical, has long been established as reliable; and this therefore gives rise to a presumption in favour of other branches, even though these might be of a completely different nature. Besides, when you are outside the realm of experience, you can be certain that you will not be refuted by experience. The urge to extend our knowledge is so great, that we can be stopped in our tracks only by running up against a clear contradiction. However, this can be avoided simply by formulating our fictions with care — but they are still fictions none the less.
<i17> Mathematics gives us a shining example of how far we can get in apriori knowledge independently of experience. Now, mathematics deals with objects and knowledge only so far as they are capable of being displayed in intuition. But this circumstance is easily overlooked, because the said intuition can itself be given apriori, and hence is hardly distinguishable from a pure apriori concept. Captivated by such a proof of the power of reason, our drive to extend our knowledge knows no limits.
<i18> The light dove, cleaving the air in free flight and feeling its resistance, could get the idea that it would be much better off in empty space. [B9] In the same way, Plato abandoned the world of the senses as imposing excessively narrow restrictions on the understanding, and ventured beyond them on the wings of his Ideas in the empty space of the pure understanding. He did not realise that, despite all his efforts, he had made no progress, since there was no resistance to serve as a sort of launchpad against which he could brace himself, and exert his forces in order to set his understanding in motion.
<i19> However, when indulging in speculation, it is the normal fate of human reason to complete the structure it is building as quickly as possible, and only afterwards to investigate whether its foundations were properly laid. But at that point all sorts of excuses will be sought to reassure us of its strength, or, even better, to persuade us against such a delayed and risky examination. During the building of the structure, what keeps us free from any worry or suspicion, and soothes us with its illusory solidity, is the following. A large part — perhaps the largest part — of the business of our reason consists in the analysis of the concepts which we already have of objects. This provides us with a considerable amount of knowledge. But in fact this knowledge is nothing more than a clarification or explanation of what we have already thought in our concepts, if only in a confused way. Nevertheless, at least as far as its form is concerned, this knowledge is valued as if it consisted in new insights, even though, as far as its matter or content is concerned, it does not extend the concepts we already have, but merely elucidates them.
<i20> [B10] Since this procedure does result in genuine apriori knowledge which constitutes a reliable and useful advance, our reason is unconsciously deluded into smuggling in assertions of a very different sort. In these assertions, our reason adds entirely alien concepts to those it already has (and indeed it does so apriori), but without our knowing how it achieves this, and without our letting any such question even enter our thoughts. So the very next topic I shall deal with is the distinction between these two kinds of knowledge.
<i21> 4. On the distinction between analytic and synthetic judgments
(In what follows, I shall consider only affirmative judgments, since what I say can easily be extended to negative ones.) In all judgments in which what is thought is the relation of a subject to its predicate, this relation can be of two sorts:
In the first case I call the judgment analytic, and in the second case I call it synthetic. Thus analytic judgments (affirmative ones, that is) are those in which the connection between the predicate and the subject is thought through identity, whereas those in which this connection is thought without identity are to be called ‘synthetic’ judgments.
<i22> [B11] Analytic judgments could also be called ‘clarificatory’, and synthetic judgments ‘expansive’. An analytic judgment does not use the predicate to add anything to the concept of the subject, but merely breaks up the concept of the subject into its component sub-concepts, which were already thought in it, if in a confused way. By contrast, a synthetic judgment adds to the concept of the subject a predicate which was not thought in it at all, and which could not have been extracted from it by any process of breaking it up into its constituent parts.
<i23> For example, if I say ‘All bodies are extended,’ this is an analytic judgment, since I need not go beyond the concept which I attach to the word ‘body’, in order to find that extension is connected to it. To extract the predicate of extension from the concept of body, I merely have to break it up into its constituent parts — in other words, I only need to become conscious of the complex whole which I always think in the concept, in order to encounter this predicate in it. So it is an analytic judgment. By contrast, if I say ‘All bodies are heavy,’ the predicate is something quite different from what I think in the mere concept of a body in general. So the addition of such a predicate to the concept of the subject results in a synthetic judgment.
<i24> Judgments of experience are, as such, all synthetic. It would be absurd to base an analytic judgment on experience, because I have no need at all to go beyond my concept in order to form the judgment, and thus the evidence of experience is unnecessary for it. That a body is extended is a proposition which is apriori certain, and not a judgment of experience. [B12] Rather than appealing to experience, I already have all the prerequisites for my judgment in the concept. I only have to draw the predicate out of it in accordance with the law of contradiction, and in doing so I can at the same time become conscious of the necessity of the judgment — something that experience would never have taught me.
<i25> By contrast, I do not include the predicate ‘heavy’ in the concept of a body in general. However, the concept of a body does denote an object of experience through one of the components of an experience; and I can add to it another component of the same experience as belonging to the first component, by virtue of which it is a body. I can know the concept of body analytically, in advance of experience, through the characteristics of extension, impenetrability, shape, etc., all of which are thought in this concept. Now I expand my knowledge by reviewing the experience from which I abstracted this concept of body, and, finding that heaviness is always connected with the above characteristics, I add the predicate ‘heavy’ to the concept of body synthetically. So it is experience which grounds the possibility of the synthesis of the predicate ‘heavy’ with the concept of body. Although neither of the two concepts is contained in the other, they are nevertheless parts of a whole, namely experience, which is itself a synthetic combination of intuitions. So they belong to each other through experience, even if only contingently.
<i26> However, the resource of experience is entirely lacking in the case of synthetic apriori judgments. If I am to go beyond a concept A, [B13] so as to know that a concept B is connected with it, what is there that will justify me in making this connection? And what is there that will make the synthesis possible, given that in this case I do not have the opportunity to look for it in the world of experience? Take the proposition ‘Everything which happens has its cause.’ In the concept of something which happens, I think an existence before which there is a time, etc., and analytic judgments can be derived from this. But the concept of a cause lies entirely outside that concept. It denotes something different from that which happens, and therefore is certainly not contained in the latter representation.
<i27> So how do I get to say about that which happens in general, something which is completely different from it? How can I know that the concept of cause belongs to it, and even belongs to it necessarily, despite the fact that the concept of cause is not contained in the concept of that which happens? What, here, is the unknown=x which the understanding relies on, when it believes it can discover, outside the concept of A, a predicate B which is alien to A, but which the understanding considers to be connected with it? It cannot be experience, because the axiom appealed to connects the two representations together, not only with greater generality than experience can provide, but also with the stamp of necessity, and hence completely apriori and on the basis of concepts alone.
<i28> The whole aim of our apriori speculative knowledge depends on synthetic axioms like this, which extend our knowledge. Analytic judgments are indeed very important and necessary, but only [B14] in order to attain the clarity of concepts which is required for a reliable synthesis which extends our knowledge as a genuinely new acquisition.
<i29> 5. All theoretical sciences of reason contain apriori synthetic judgments as principles
1. Mathematical judgments are all synthetic. Up to now, this proposition seems to have escaped the notice of analysts of human understanding, and even seems to directly contradict everything they have assumed. Nevertheless, it is incontrovertibly certain, and very important in its consequences. These analysts found that all mathematical reasoning proceeds in accordance with the principle of contradiction, as is required by the nature of demonstrative certainty. They therefore persuaded themselves that the axioms of mathematics could also be derived from the principle of contradiction. But they were wrong in this. It is true that a synthetic proposition can always be assessed in accordance with the principle of contradiction, but only as presupposing another synthetic proposition from which it can be deduced. Synthetic propositions can never be self-evident.
<i30> First it must be observed that all mathematical judgments are essentially apriori and not empirical, because they involve necessity, which can never be derived from experience. [B15] But if this is not accepted, then I am willing to limit my proposition to pure mathematics, since its very concept implies that it does not include any empirical knowledge, but only pure apriori knowledge.
<i31> At first sight, it might be thought that the proposition 7+5=12 is merely an analytic proposition, which follows from the concept of a sum of seven and five in accordance with the law of contradiction. But if you consider it more closely, you will find that the concept of the sum of seven and five contains no more than the union of the two numbers in a single number. Through this concept, there is no thought at all of what the single number is which embraces both the others. In no way do I already think the concept of twelve simply by virtue of my thinking the union of seven and five, and however long I analyse my concept of such a possible sum, I shall not find twelve in it.
<i32> In order to find the number twelve, I have to go beyond the above concepts, and obtain the help of an intuition which corresponds to one of the two concepts. This might be the five fingers of my hand, or five points (following Segner in his Arithmetic). I then add, one by one, the units of the five which is given in the intuition, to the concept of seven. For I first take the number seven, and then take the fingers of my hand as an intuition which helps me represent the concept of five. One by one, I add to the number seven the units which I previously took as a single whole [B16] making up the number five. So it is with the aid of the image of my hand that I see the number twelve emerge. In the concept of the sum = 7+5, I did indeed think that 5 must be added to 7, but not that this sum is equal to the number twelve.
<i33> Therefore arithmetical propositions are always synthetic. This is even clearer in the case of somewhat larger numbers, since it is obvious that, however we twist and turn our concepts, we could never find the sum of the numbers by the mere analysis of our concepts, and without getting the help of intuition.
<i34> Just as little is any axiom of pure geometry analytic. That the straight line between two points is the shortest is a synthetic proposition, since my concept of straight contains nothing about quantity, but only a quality. Therefore the concept of the shortest is entirely additional to it, and cannot be derived by any analysis from the concept of a straight line. So here the help of intuition must be obtained, since it is only through the mediation of intuition that the synthesis of the concepts is possible.
<i35> A few axioms presupposed by geometry are indeed genuinely analytic and depend on the principle of contradiction. However, like identical propositions, their only function is as links in the deductive chain, and [B17] not as axioms — for example, a=a, i.e. the whole is similar to itself, or (a+b)>a, i.e. the whole is greater than its part. And yet, although these are valid by virtue of concepts alone, they too are allowed in mathematics only because they can be displayed in intuition.
<i36> Here, what makes us usually believe that the predicate of such demonstratively certain judgments already lies in our concept of the subject, and therefore that the judgment is analytic, is merely an ambiguity of expression. It is said that we ought to join in thought a particular predicate to a given concept, and that this necessity is already inherent in the concepts. But the question is not what we ought to add in thought to the given concept, but what we actually think in it, however obscurely. Then it is obvious that the predicate does indeed necessarily belong to the given concept, but not as being thought in the concept itself, but through the mediation of an intuition which must be added to the concept.
<i37> 2. Natural science (physics) contains within itself synthetic apriori judgments as axioms. I shall mention just a couple of axioms as examples:
It is clear that both axioms are not only necessary (and hence have an apriori origin), but are also synthetic [B18] propositions. For I do not think permanence in the concept of matter, but merely its presence in the space it fills. In fact I have to go beyond the concept of matter, in order to add to it apriori in thought something I did not think in it. And the same goes for the remaining propositions of the pure part of natural science.
<i38> 3. As for metaphysics, you may consider it to be a science which has not yet been established, but which the nature of human reason makes unavoidable. Even if this is your view, metaphysics must consist of synthetic apriori knowledge. It is not its business merely to analyse concepts of things which we form for ourselves apriori, and thus to clarify them analytically. Rather, our ambition is to extend our apriori knowledge. To do this we must use axioms which add to the given concept something which was not already contained in it, and to take synthetic apriori judgments so far that experience itself cannot follow us — for example, in the proposition: ‘The world must have a first beginning,’ and the such like. Thus metaphysics, at least as far as its objective is concerned, consists of nothing but synthetic apriori propositions.
<i39> [B19] 6. The general problem of pure reason
We have already made great progress if we can bring a number of investigations under the formula of a single problem. In this way we not only lighten our own task by defining it precisely, but we also make it easier for others to assess whether or not we have successfully met our objectives. The essential problem of pure reason is contained in the question:
<i40> The reason why metaphysics has till now remained in such a precarious state of uncertainty and controversy is entirely because no-one has previously let this problem enter their minds — or perhaps even the distinction between analytic and synthetic judgments. Metaphysics now stands or falls on the solution of this problem, or on a satisfactory proof that in fact there is no such possibility, and hence nothing which requires an explanation.
<i41> Among all philosophers, David Hume came nearest to dealing with this problem; but he fell far short of defining it closely enough, or of conceiving it in its full universality. Instead, he confined himself to the synthetic proposition about the connection between an effect and its cause (the principle of causality), and he believed [B20] he had shown that such an apriori proposition is completely impossible. Following his arguments, everything we call metaphysics would amount to no more than a mere delusion, in which we suppose ourselves to have rational insight into what is in fact merely borrowed from experience, and has taken on the appearance of necessity as the result of habitual repetition. This claim is destructive of all pure philosophy, and Hume would never have stooped to it if he had kept our problem fixed before his eyes in its full universality. He would have recognised that, according to his argument, there could be no such thing as pure mathematics either, since it certainly consists of synthetic apriori propositions. But his good sense would surely have kept him from making any such claim.
<i42> In solving the above problem, we are at the same time dealing with the possibility of the pure use of reason for grounding and developing all sciences which contain theoretical apriori knowledge of objects — that is, we are answering the questions:
Since these sciences are actually given, it is perfectly proper to ask how they are possible, since their actuality proves that they must be possible.*
[<i43> *Some people could have doubts as to whether this last point is true of natural science. But you only need to consider the various propositions which come at the very beginning of physics proper (empirical physics), such as those about the permanence of the same quantity of matter, inertia, the equality of action and reaction, etc. You will soon be convinced that they constitute a pure or rational physics, which fully deserves to be established separately as a science in its own right, and with its own exclusive territory, whether this is strictly limited or more extensive.]
<i44> As for metaphysics, however, [B21] everyone must be left with grounds for doubting its possibility, both because of its poor progress so far, and because no single metaphysical system put forward up till now can be described as actually established, at least as far as the essential goal of metaphysics is concerned.
<i45> However, there is another sense in which this kind of knowledge can also be considered as given. Even if metaphysics has not been actualised as a science, it is actual as a natural disposition (natural metaphysics). For, quite apart from being motivated by simple pride in wealth of knowledge, human reason advances unstoppably, driven by its own special urge, to questions which cannot be answered through any use of reason within the world of experience, or through principles borrowed from experience. So a kind of metaphysics always has been, and always will be actualised in every human being, as soon as their reason extends to speculation. So this now raises another question, namely: [B22]
That is, how does the nature of universal human reason give rise to questions which pure reason poses for itself, and which it is driven, by its own special urge, to answer as well as it can?
<i46> All previous attempts to answer these natural questions (such as whether the world has a beginning, or has existed since eternity) have always led to unavoidable contradictions. So the task cannot be left simply to the natural disposition towards metaphysics — that is, to the pure faculty of reason itself, which permanently gives rise to some sort of metaphysics, whatever it may be. It must be possible for reason to attain certainty as to whether the objects of metaphysics are knowable or not — that is, to come to a decision, either about the objects of its questions, or about the capacity or incapacity of reason to make judgments about them. If so, we can either extend the scope of our pure reason with confidence, or set definite and firm limits to it. The final question, which arises out of the general problem above, would justifiably be the following:
Thus the critique of reason necessarily leads in the end to scientific knowledge. By contrast, the dogmatic use of reason without a critique leads to unfounded assertions, [B23] which can be counterbalanced by equally plausible opposite ones — and hence leads to scepticism.
<i47> We should not be deterred from embarking on this science on the grounds that it will be an enormous undertaking, since it does not deal with the objects of reason, which are indeed infinitely various. Rather, it merely deals with itself — with problems which arise entirely from within itself, and are posed for it by its own nature, and not by the nature of things distinct from it. Once reason has become completely familiar with its own powers in relation to the objects it might encounter in experience, it must become easy for it to determine, with completeness and certainty, the extent and limits of its attempted use beyond all bounds of experience.
<i48> So we can and must consider as abortive all the attempts which have been made till now to establish a metaphysics dogmatically. Whatever analytic content the various systems have consists merely in the dissection of the concepts which reside apriori in our reason. This is not the goal of metaphysics at all, but only a preparation for metaphysics proper, the goal of which is to extend our synthetic apriori knowledge. Analysis is useless for achieving this goal, because it merely shows what is contained in these concepts. It does not show how we attain such concepts apriori, so that we can then also determine their valid use in relation to the objects [B24] of all knowledge in general.
<i49> It takes little self-denial to give up all these dogmatic claims, since the authority of every metaphysical system so far has long since been destroyed by the internal contradictions of reason with itself. These cannot be covered up, and they are also unavoidable if reason proceeds dogmatically. More perseverance will be required not to be held back by the intrinsic difficulty of the task at hand, or by resistance from others. This task is to apply a method which is diametrically opposed to any that has been used before, in order finally to bring about a thriving and fruitful growth in a science which human reason cannot avoid. You can cut off every stem it has sprouted till now, but you cannot dig out its roots.
<i50> 7. The idea and structure of a distinct science called ‘the critique of pure reason’
All I have said so far results in the idea of a distinct science which can be called the critique of pure reason. Reason is the faculty which makes the principles of apriori knowledge ready for use. Hence it is pure reason that contains the principles by which we can know something strictly apriori. An instrument of pure reason would be the totality of principles by which all [B25] pure apriori knowledge can be acquired and actualised. The complete application of such an instrument would result in a system of pure reason. However, it would be overambitious to expect such a system at this stage, when it is still uncertain whether an extension of our knowledge in this area is even possible at all, and if it is, in which cases. So we can regard a science which merely assesses pure reason, its sources, and its limits, as an introduction to the system of pure reason. Such an introduction would not amount to a body of knowledge to be taught. Rather, it must be no more than a critique of pure reason. Its usefulness would in fact be purely negative as far as speculation is concerned, since it would serve, not to widen the scope of our reason, but only to purge it, and keep it free of errors — but this is already a considerable achievement.
<i51> I call ‘transcendental’ all knowledge which deals universally, not so much with objects, but with our way of knowing objects, in so far as this is to be possible apriori. A system of such concepts would be called ‘transcendental philosophy’. But again, even this is too much for the initial stage. Such a science would have to cover the whole of analytic knowledge as well as of synthetic apriori knowledge. So its scope is too wide for our purposes, since we must carry analysis no further than is absolutely necessary for us to understand, in their whole extent, the principles of apriori synthesis, which are our only concern here.
<i52> [B26] It is this investigation with which we are now concerned. We cannot call it a body of knowledge to be taught, but only a transcendental critique, since its purpose is not to extend knowledge as such, but only to correct it, and to provide the touchstone for whether any apriori knowledge is valid or not. So such a critique is a preparation, as far as is possible, for an instrument of pure reason; and if this should turn out not to be possible, then at least for a rule-book of pure reason. This rule book would show how it might be possible, at some time in the future, to lay out the complete system of the philosophy of pure reason, using both the analytic and the synthetic methods — irrespective of whether it extends or merely limits the knowledge pure reason can attain.
<i53> Even in advance it can be calculated that such a system is possible, and indeed that it could not be too great in scope for us to hope to complete it entirely. This is because our subject-matter is not the nature of things, which is inexhaustible, but the understanding, which makes judgments about the nature of things — and then only in relation to its apriori knowledge. Since we do not need to look for its stock of apriori knowledge in sources external to us, it cannot remain hidden from us, and it is likely to be small enough for it to be catalogued in its entirety, assessed as to its validity or otherwise, and given its correct valuation.
<i54> [B27] Still less should you expect here a critique of books or systems of pure reason, but rather a critique of the pure faculty of reason itself. It is only on the basis of the critique of pure reason that we have a reliable touchstone for evaluating the philosophical content of old and new works in this discipline. Without a critique, unqualified historians and critics judge the groundless assertions of others in the light of their own equally groundless assertions.
<i55> Transcendental philosophy is the idea of a science, for which the critique of pure reason has to lay out the whole plan architectonically — that is, on the basis of principles, with an absolute guarantee of the completeness and robustness of all the parts which make up this construction. Transcendental philosophy is the system of all the principles of pure reason. However, this critique does not yet amount to a transcendental philosophy in the full sense of the term, simply because it is not a complete system unless it includes an exhaustive analysis of all human apriori knowledge.
<i56> Our critique must certainly provide us with a complete list of all the foundational concepts on which pure human knowledge is based. However, it stops short of a detailed analysis of these concepts themselves, let alone a complete review of all the concepts which can be derived from them. This is reasonable on two counts:
It will be easy to complete the analysis of the foundational concepts and the subsequent derivation of the apriori concepts they give rise to, once they are established as a complete list of the principles of synthesis, and nothing belonging to their essential purpose is left out.
<i57> So the critique of pure reason covers everything which belongs to transcendental philosophy. But although it constitutes the complete idea of transcendental philosophy, it is not this science itself, since it does not pursue analysis beyond what is necessary for the complete assessment of synthetic apriori knowledge.
<i58> In specifying the sub-divisions of such a science, the most important thing to pay attention to is that it must under no circumstances admit any concepts which have any empirical content whatever. In other words, the knowledge must be utterly pure apriori. It follows that, although the most fundamental axioms and concepts of morality are instances of apriori knowledge, they do not belong to transcendental philosophy. It is true that the concepts [B29] of pleasure and pain, appetite, inclination, and so on, which are all of empirical origin, themselves play no role in the basic prescriptions of morality. Nevertheless, they must necessarily be included in the drawing up of a system of pure morality through the concept of a duty, whether as an obstacle to be overcome, or as an impulse which should not be turned into a reason for acting. So transcendental philosophy is confined to pure speculative reason. Since everything practical involves motives, it relates to feelings, which belong to empirical sources of knowledge.
<i59> If we want to set up the division of the science I am now proposing on the general lines of any system whatever, then it must include first a Doctrine of Elements, and second a Doctrine of Method. Each of these main divisions will have its sub-divisions; but I am not yet in a position to explain the grounds for these. All I need to say by way of introduction or preliminary reminder, is that there are two branches of human knowledge, which perhaps grow from a common root, even if it is unknown to us. These are sensibility and understanding. Through the former objects are given to us, and through the latter they are thought. Sensibility will belong to transcendental philosophy in so far as it contains apriori representations which constitute the preconditions for [B30] objects to be given to us. The transcendental doctrine of sensibility must belong to the first part of the science of elements, because the preconditions for objects to be given to human knowledge precede the preconditions for these objects to be thought.