The natural instinct of a man in the water is to do precisely that which will infallibly cause him to be drowned; learning to swim means acquiring the habit of suppressing spontaneous movements and performing others instead. Similarly, criticism is not a natural habit; it must be inculcated, and only becomes organic by dint of continued practice. Historical work is, then, pre-eminently critical; whoever enters upon it without having first been put on his guard against his instinct is sure to be drowned in it. In order to appreciate the danger it is well to examine one's conscience and analyse the causes of that _ignavia_ which must be fought against till it is replaced by a critical attitude of mind.[63] It is also very salutary to familiarise oneself with the principles of historical method, and to analyse the theory of them, one by one, as we propose to do in the present volume. "History, like every other study, is chiefly subject to errors of fact arising from inattention, but it is more exposed than any other study to errors due to that mental confusion which produces incomplete analyses and fallacious reasonings.... Historians would advance fewer affirmations without proof if they had to analyse each one of their affirmations; they would commit themselves to fewer false principles if they made it a rule to formulate all their principles; they would be guilty of fewer fallacies if they were obliged to set out all their arguments in logical form."[64] _SECTION I.--EXTERNAL CRITICISM_ CHAPTER II TEXTUAL CRITICISM Let us suppose that an author of our own day has written a book: he sends his manuscript to the printer; with his own hand he corrects the proofs, and marks them "Press." A book which is printed under these conditions comes into our hands in what is, for a document, a very good condition. Whoever the author may be, and whatever his sentiments and intentions, we can be certain--and this is the only point that concerns us at present--that we have before us a fairly accurate reproduction of the text which he wrote. We are obliged to say "fairly accurate," for if the author has corrected his proofs badly, or if the printers have not paid proper attention to his corrections, the reproduction of the original text is imperfect, even in this specially favourable case. Printers not unfrequently make a man say something which he never meant to say, and which he does not notice till too late.