The problem that metaphysical philosophy confronted, as Kant illustrates, was its aspirations to establish a world of immutable truths, which is an enterprise that can be conducted successfully in the analytic sciences, logic and so forth (cf.Aristotelian logic), but has had little success in the synthetic sciences, metaphysics.274 So in order to extend the reach of enumerative induction from the merely relative (which entails that Aristotle’s authority must be questioned) to the absolute, metaphysical philosophy was forced to introduce a transcendent argument (albeit already implied by Aristotle) into epistemology, whereby enumerative induction was suddenly endowed with the ability to establish and reachabsolute truths about objects, a synthetic a priori.275 In addition, the introduction of the transcendent argument entails that it must be possible to deduce (transcendent) absolute truths frominductively established premisses. Necessary truths can thus be derived from contingent truths. But one is still confronted with the fact that the conclusions drawn from the synthetic a priori, which itself is established a posteriori, can only be necessarily true in a relative, contingent meaning provided that facts have not changed since the last observation. Hence, the validity of the synthetic a priori is contingent rather than apodictic. What is problematic to classical metaphysical philosophy is the fact that its demonstrative method of reasoning is circular, leaving the real relationship between induction and deduction undetermined, as induction itself must lead to necessary truths, axioms, which in turn make up the premisses from which it is possible to demonstrate additional necessary consequences and truths.276 Yet, science, the art of demonstration, demanded that each and every argument, analytical and synthetic alike, expresp a r t i 1 , c h a p t e r 4 124 4 . 2 . 2 the transcendent argument 274 Kant, Cr. P. R., pp.A vii-ix, B vii-viii, B x-xi, and B xiv-xv. 275 Cf., e.g., ibid., pp.A103-110. 276 Wedberg, Historia 2, pp. 39-80.