The refutation of Trendelenburg’s argument which is offered by Caird[468] is inconclusive. Caird assumes the chief point at issue, first by ignoring the possibility that space may be known _a priori_ in reference to appearances and yet at the same time be transcendently real; and secondly by ignoring the fact that to deny spatial properties to things in themselves is as great a violation of Critical principles as to assert them. One point, however, in Caird’s reply to Trendelenburg calls for special consideration, viz. Caird’s contention that Kant did actually take account of the third alternative, rejecting it as involving the “absurd” hypothesis of a pre-established harmony.[469] Undoubtedly Kant did so. But the contention has no relevancy to the point before us. The doctrine of pre-established harmony is a metaphysical theory which presupposes the possibility of gaining knowledge of things in themselves. For that reason alone Kant was bound to reject it. A metaphysical proof of the validity of metaphysical judgments is, from the Critical point of view, a contradiction in terms. As the validity of _all_ speculations is in doubt, a proof which is speculative cannot meet our difficulties. And also, as Kant himself further points out, the pre-established harmony, even if granted, can afford no solution of the Critical problem how _a priori_ judgments can be passed upon the independently real. The judgments, thus guaranteed, could only possess _de facto_ validity; we could never be assured of their necessity.[470] It is chiefly in these two inabilities that Kant locates the “absurdity” of a theory of pre-established harmony. The refutation of that theory does not, therefore, amount to a disproof of the possibility which we are here considering.