said, 'this I was led to infer') 'by his sometimes introducing a word which was wholly out of place--_entertained_ instead of _denied_, for example--but which I presently found to belong to the next sentence, perhaps four or five lines further on, which he had been preparing at the very moment when he gave me the words of the one that preceded it.' In the same way I have often unconsciously substituted one word for another in lecturing, the word used always belonging to a later sentence than the word intended to be used. I have noticed also this peculiarity, that when a substitution of this kind has been once made, an effort is required to avoid repeating the mistake, even if it be not repeated quite unconsciously to the end of the discourse. In this way, for example, I once throughout an entire lecture used the word 'heavens' for the word 'screen' (the screen on which lantern pictures were shown). A similar peculiarity may be noticed with written errors. Thus in my treatise on a scientific subject, in which the utmost care had been given to minute points of detail, I once wrote 'seconds' for 'minutes' throughout several pages--in fact, from the place where first the error was made, to the end of the chapter. (See the _first_ edition of my _Transits of Venus_, pp. 131-136, noting as an additional peculiarity that the whole object of the chapter in which this mistake was made was to show how many minutes of difference existed between the occurrence of certain events.) An even more curious instance of a mistake arising from doing one thing while thinking of another occurred to me fourteen years ago. I was correcting the proof-sheets of an astronomical treatise in which occurred these words: 'Calling the mean distance of the earth 1, Saturn's mean distance is 9·539; again, calling the earth's period 1, Saturn's mean period is 29·457:--now what relation exists between these numbers 9·539 and 29·457 and their powers? The first is less than the second, but the square of the first is plainly greater than the second; we must therefore try higher powers, &c. &c.' The passage was quite correct as it stood, and if the two processes by which I was correcting verbal errors and following the sense of the passage had been really continuous processes of thought, unquestionably the passage would have been left alone. If the passage had been erroneous and had been simply left in that condition the case would have been one only too familiar to those who have had occasion to correct proofs. But what I actually did was deliberately to make nonsense of the passage while improving the sound of the second sentence. I made it run, 'the first is less than the second, but the square of the first is plainly greater than the square of the second,' the absurdity of which statement a child would detect. If the first proof in its correct form, with the incorrect correction carefully written down in the margin,