If we are never to be certain, after having been once certain wrongly, then we ought never to attempt a proof because we have once made a bad one. Errors in reasoning are lessons and warnings, not to give up reasoning, but to reason with greater caution. It is absurd to break up the whole structure of our knowledge, which is the glory of the human intellect, because the intellect is not infallible in its conclusions. If in any particular case we have been mistaken in our inferences and the certitudes which followed upon them, we are bound of course to take the fact of this mistake into account, in making up our minds on any new question, before we proceed to decide upon it. But if, while weighing the arguments on one side and the other and drawing our conclusion, that old mistake has already been allowed for, or has been, to use a familiar mode of speaking, discounted, then it has no outstanding claim against our acceptance of that conclusion, after it has actually been drawn. Whatever be the legitimate weight of the fact of that mistake in our inquiry, justice has been done to it, before we have allowed ourselves to be certain again. Suppose I am walking out in the moonlight, and see dimly the outlines of some figure among the trees;—it is a man. I draw nearer,—it is still a man; nearer still, and all hesitation is at an end,—I am certain it is a man. But he neither moves, nor speaks when I address him; and then I ask myself what can be his purpose in hiding among the trees at such an hour. I come quite close to him, and put out my arm. Then I find for certain that what I took for a man is but a singular shadow, formed by the falling of the moonlight on the interstices of some branches or their foliage. Am I not to indulge my second certitude, because I was wrong in my first? does not any objection, which lies against my second from the failure of my first, fade away before the evidence on which my second is founded?