of any scientific theory always imply that that alone can be real which is coherent, which forms part of an intelligible system. To say that thinking is the test of reality may seem to open up the way to the most mischievous and unscientific delusions of metaphysics: metaphysicians being supposed to be persons who evolve the world out of their inner consciousness, instead of making their minds the passive mirrors of reality (whatever that may mean). But we are familiar with this test of reality in its negative form — the inconceivability of the opposite. This test has sometimes been discredited for two reasons: —
(1) Conceiving has been taken to mean representing in a mental image or picture, whereas it is only in the sense in which conceiving means thinking that inconceivability can be the test of truth. (2) We are very apt to suppose we can or cannot think something, simply because we have not taken all the conditions into account. Thus, (1) when the infinity of time or space is discussed, our incapacity to form a mental picture of infinite time or space has been taken as if it were a consideration that weighed against our incapacity to think a limit in time or in space without contradiction. (2) People used to think the Antipodes inconceivable, because they thought of gravity as a force acting in the direction of an absolute down: that human beings, constituted in any such way that we could consider them human beings, should be able to walk on the lower side of the world meant, to the disbelievers in the Antipodes, the same sort of thing as if it were said that we here could walk like flies on the inside of the roof with our heads down. Change the meaning of gravitation, change the meaning of up and down, and it becomes inconceivable that a man walking in New Zealand should fall off into air, because falling off would mean falling up, which is a contradiction. The obvious difficulty of applying the test safely comes simply from the difficulty of being sure that we have exhausted all the relevant conditions. And that is why we can only apply the test easily in very abstract matters, where we have purposely eliminated all except the very simplest conditions, e.g. in the mathematical sciences. In the case of
