PLACE OF HYPOTHESIS IN EXPERIMENTAL SCIENCE. 559

mind in explaining phenomena there must be, is clear ; the difficulty is in proving that it has been reached. Nor, if we suppose that in certain fields it has been reached, does that fact warrant us in assuming the absolute truth of the hypo- thesis in question. It may not be possible for us to attain absolute truth by means of hypotheses. Mill distinguished between hypotheses which rest on mere analogy and such as are capable of " being ultimately brought to the test of actual induction," claiming for the latter a verifiability which he denied to the former (Logic, 8th ed., vol. ii., pp. 15, 94). And on this ground he refused the title of " positive truths " to both the emission and the undulatory theory of light. In fact, however, neither of these theories stands in a worse position than the commonplace hypothesis of an atmosphere. The atmosphere is just as little capable of being brought to the test of actual induction as is the luminiferous ether. Its existence is assumed upon the ana- logy of watery fluid for the purpose of accounting for certain phenomena, just as is the existence of the ether. We are so familiar with the hypotheses of an atmospheric medium that we have ceased to regard it as an hypothesis, and it requires a certain effort of thought to realise that it is so ; yet nothing can be more certain than that if we had no ex- perience of visible fluid, however we might explain the motions of the clouds and the pressure of the air upon us, it could not be by means of an atmosphere. We should have no experience capable of yielding the notion. It is clear therefore that the atmosphere, as distinct from the percep- tions for which it accounts, is not, strictly speaking, a " real phenomenon " ; is not, in fact, a sensible, but a mere hypo- thesis. It is possible that at some future time the idea of a luminiferous ether may be so familiar that its hypothetical character may be forgotten. In any case it is impossible to draw any valid distinction between the two hypotheses, except so far as the hypothesis of an ethereal medium may not have been so thoroughly tested as the hypothesis of an atmospheric medium has been. The latter hypothesis is, however, accepted for no other reason than that it explains certain facts, and enables us to interpret them as the result of undulations of the said medium ; and until men shall have acquired the porcine faculty of seeing the wind it will remain entirely " unsusceptible of being brought to the test of actual induction ". It follows that the theory of sound, which is based upon the hypothesis of an aerial medium, is not empirically verifi- able. Nay the entire molecular theory is incapable of em-