The cosmological principle of totality could not give us any certain knowledge in regard to the maximum in the series of conditions in the world of sense, considered as a thing in itself. The actual regress in the series is the only means of approaching this maximum. This principle of pure reason, therefore, may still be considered as valid- not as an axiom enabling us to cogitate totality in the object as actual, but as a problem for the understanding, which requires it to institute and to continue, in conformity with the idea of totality in the mind, the regress in the series of the conditions of a given conditioned. For in the world of sense, that is, in space and time, every condition which we discover in our investigation of phenomena is itself conditioned; because sensuous objects are not things in themselves (in which case an absolutely unconditioned might be reached in the progress of cognition), but are merely empirical representations the conditions of which must always be found in intuition. The principle of reason is therefore properly a mere rule- prescribing a regress in the series of conditions for given phenomena, and prohibiting any pause or rest on an absolutely unconditioned. It is, therefore, not a principle of the possibility of experience or of the empirical cognition of sensuous objects- consequently not a principle of the understanding; for every experience is confined within certain proper limits determined by the given intuition. Still less is it a constitutive principle of reason authorizing us to extend our conception of the sensuous world beyond all possible experience. It is merely a principle for the enlargement and extension of experience as far as is possible for human faculties. It forbids us to consider any empirical limits as absolute. It is, hence, a principle of reason, which, as a rule, dictates how we ought to proceed in our empirical regress, but is unable to anticipate or indicate prior to the empirical regress what is given in the object itself. I have termed it for this reason a regulative principle of reason; while the principle of the absolute totality of the series of conditions, as existing in itself and given in the object, is a constitutive cosmological principle. This distinction will at once demonstrate the falsehood of the constitutive principle, and prevent us from attributing (by a transcendental subreptio) objective reality to an idea, which is valid only as a rule.
In order to understand the proper meaning of this rule of pure reason, we must notice first that it cannot tell us what the object is, but only how the empirical regress is to be proceeded with in order to attain to the complete conception of the object. If it gave us any information in respect to the former statement, it would be a constitutive principle- a principle impossible from the nature of pure reason. It will not therefore enable us to establish any such conclusions as: "The series of conditions for a given conditioned is in itself finite." or, "It is infinite." For, in this case, we should be cogitating in the mere idea of absolute totality, an object which is not and cannot be given in experience; inasmuch as we should be attributing a reality objective and independent of the empirical synthesis, to a series of phenomena. This idea of reason cannot then be regarded as valid- except as a rule for the regressive synthesis in the series of conditions, according to which we must proceed from the conditioned, through all intermediate and subordinate conditions, up to the unconditioned; although this goal is unattained and unattainable. For the absolutely unconditioned cannot be discovered in the sphere of experience.
We now proceed to determine clearly our notion of a synthesis which can never be complete. There are two terms commonly employed for this purpose. These terms are regarded as expressions of different and distinguishable notions, although the ground of the distinction has never been clearly exposed. The term employed by the mathematicians is progressus in infinitum. The philosophers prefer the expression progressus in indefinitum. Without detaining the reader with an examination of the reasons for such a distinction, or with remarks on the right or wrong use of the terms, I shall endeavour clearly to determine these conceptions, so far as is necessary for the purpose in this Critique.
We may, with propriety, say of a straight line, that it may be produced to infinity. In this case the distinction between a progressus in infinitum and a progressus in indefinitum is a mere piece of subtlety. For, although when we say, "Produce a straight line," it is more correct to say in indefinitum than in infinitum; because the former means, "Produce it as far as you please," the second, "You must not cease to produce it"; the expression in infinitum is, when we are speaking of the power to do it, perfectly correct, for we can always make it longer if we please- on to infinity. And this remark holds good in all cases, when we speak of a progressus, that is, an advancement from the condition to the conditioned; this possible advancement always proceeds to infinity. We may proceed from a given pair in the descending line of generation from father to son, and cogitate a never-ending line of descendants from it. For in such a case reason does not demand absolute totality in the series, because it does not presuppose it as a condition and as given (datum), but merely as conditioned, and as capable of being given (dabile).
Very different is the case with the problem: "How far the regress, which ascends from the given conditioned to the conditions, must extend"; whether I can say: "It is a regress in infinitum," or only "in indefinitum"; and whether, for example, setting out from the human beings at present alive in the world, I may ascend in the series of their ancestors, in infinitum- mr whether all that can be said is, that so far as I have proceeded, I have discovered no empirical ground for considering the series limited, so that I am justified, and indeed, compelled to search for ancestors still further back, although I am not obliged by the idea of reason to presuppose them.
My answer to this question is: "If the series is given in empirical intuition as a whole, the regress in the series of its internal conditions proceeds in infinitum; but, if only one member of the series is given, from which the regress is to proceed to absolute totality, the regress is possible only in indefinitum." For example, the division of a portion of matter given within certain limits- of a body, that is- proceeds in infinitum. For, as the condition of this whole is its part, and the condition of the part a part of the part, and so on, and as in this regress of decomposition an unconditioned indivisible member of the series of conditions is not to be found; there are no reasons or grounds in experience for stopping in the division, but, on the contrary, the more remote members of the division are actually and empirically given prior to this division. That is to say, the division proceeds to infinity. On the other hand, the series of ancestors of any given human being is not given, in its absolute totality, in any experience, and yet the regress proceeds from every genealogical member of this series to one still higher, and does not meet with any empirical limit presenting an absolutely unconditioned member of the series. But as the members of such a series are not contained in the empirical intuition of the whole, prior to the regress, this regress does not proceed to infinity, but only in indefinitum, that is, we are called upon to discover other and higher members, which are themselves always conditioned.
In neither case- the regressus in infinitum, nor the regressus in indefinitum, is the series of conditions to be considered as actually infinite in the object itself. This might be true of things in themselves, but it cannot be asserted of phenomena, which, as conditions of each other, are only given in the empirical regress itself. Hence, the question no longer is, "What is the quantity of this series of conditions in itself- is it finite or infinite?" for it is nothing in itself; but, "How is the empirical regress to be commenced, and how far ought we to proceed with it?" And here a signal distinction in the application of this rule becomes apparent. If the whole is given empirically, it is possible to recede in the series of its internal conditions to infinity. But if the whole is not given, and can only be given by and through the empirical regress, I can only say: "It is possible to infinity, to proceed to still higher conditions in the series." In the first case, I am justified in asserting that more members are empirically given in the object than I attain to in the regress (of decomposition). In the second case, I am justified only in saying, that I can always proceed further in the regress, because no member of the series. is given as absolutely conditioned, and thus a higher member is possible, and an inquiry with regard to it is necessary. In the one case it is necessary to find other members of the series, in the other it is necessary to inquire for others, inasmuch as experience presents no absolute limitation of the regress. For, either you do not possess a perception which absolutely limits your empirical regress, and in this case the regress cannot be regarded as complete; or, you do possess such a limitative perception, in which case it is not a part of your series (for that which limits must be distinct from that which is limited by it), and it is incumbent you to continue your regress up to this condition, and so on.
These remarks will be placed in their proper light by their application in the following section.