It is relatively easy to sweep away the whole of metaphysics, as Kant did. The slightest puff in its direction blows it away, and what was needed was not so much a powerful pair of lungs to provide the blast, as a powerful dose of courage to turn it against so timelessly venerable a house of cards.

But you must not think that what has then been achieved is the actual elimination of metaphysics from the empirical content of human knowledge. In fact, if we cut out all metaphysics it will be found to be vastly more difficult, indeed probably quite impossible, to give any intelligible account of even the most circumscribed area of specialisation within any specialised science you please. Metaphysics includes, amongst other things—to take just one quite crude example— the unquestioning acceptance of a more-than-physical—that is, transcendental—significance in a large number of thin sheets of wood-pulp covered with black marks such as are now before you.

Or, to take it at a deeper level: call to mind that sense of misgiving, that cold clutch of dreary emptiness which comes over everybody, I expect, when they first encounter the description given by Kirchhoff and Mach of the task of physics (or of science generally): ‘a description of the facts, with the maximum of completeness and the maximum economy of thought’; a feeling of emptiness which one cannot master, despite the emphatic and even enthusiastic agreement with which one's theoretical reason can hardly fail to accept this prescription.