the other planets were known to be subject, and to predict its place in the heavens, at any time, according to this law, we should be said to form a theory of its motions; as we should not take for granted, any new quality of a substance, or the existence of any substance, which was not evidently present, but only of _tendencies_ observed before in other circumstances,--analogous indeed, but not absolutely the same. We should be said to form an hypothesis on the subject, if, making the same prediction, as to its motions, and place in the heavens, at any given time, we were to ascribe the centripetal tendency, which confines it within its orbit, to the impulse of ether, or to any other mechanical cause. The terms, however, I must confess, though the distinction which I have now stated would be, in all cases, a very convenient one, are used very loosely, not in conversation merely, but in the writings of philosophers,--an hypothesis often meaning nothing more than a theory, to which we have not given our assent,--and a theory, an hypothesis which we have adopted, or still more, one which we have formed ourselves. A _theory_, then, even in that best sense, to which I wish it accurately confined, as often as it ventures a single hair-breadth beyond the line of former observation, may be wrong, as an hypothesis may be wrong. But, in a theory, in this sense of it, there are both less risk of error, and less extensive evil from error, than in an hypothesis. There is less risk of error, because we speak only of the properties of bodies, that must be allowed actually to exist; and the evil of error is, for the same reason, less extensive, since it must be confined to this single point; whereas, if we were to imagine falsely the presence of some third substance, our supposition might involve as many errors, as that substance has qualities; since we should be led to suppose, and expect, some or all of the other consequences, which usually attend it, when really present. The practical conclusion to be drawn from all this very long discussion, is, that we should use hypotheses to suggest and direct inquiry, not to terminate or supersede it; and that, in theorizing,--as the chance of error, in the application of a general law, diminishes, in proportion to the number of analogous cases, in which it is observed to hold,--we should not form any general proposition, till after as wide an induction, as it is possible for us to make; and, in the subsequent application of it to particulars, should never content ourselves, in any new circumstances, with the mere probability, however high, which this application of it affords; while it is possible for us to verify, or disprove it, by actual experiment. FOOTNOTES: [30] On the Powers of the Human Mind, Essay vi. Chap. viii. Vol. II. p. 334. 8vo. _edit._ [31] Traite des Systemes, chap. xii. Vol. II. p. 372. [32] Travels, Part iv, chap. 8. Swift's Works, _edit._ Nichols, Vol. ix. p. 300. LECTURE IX.