its negative criticism, necessary for the verification of the author's statements. This twofold division of the labour of criticism is, moreover, only employed by a select few. The natural tendency, even of historians who work methodically, is to read the text with the object of extracting information directly from it, without any thought of first ascertaining what exactly was in the author's mind. This procedure is excusable at most in the case of nineteenth-century documents, written by men whose language and mode of thought are familiar to us, and then only when there is not more than one possible interpretation. It becomes dangerous as soon as the author's habits of language or thought begin to differ from those of the historian who reads him, or when the meaning of the text is not obvious and indisputable. Whoever, in reading a text, is not exclusively occupied with the effort to understand it, is sure to read impressions of his own into it; he is struck by phrases or words in the document which correspond to his own ideas, or agree with his own a priori notion of the facts; unconsciously he detaches these phrases or words, and forms out of them an imaginary text which he puts in the place of the real text of the author.
II. Here, as always in history, method consists in repressing the first impulse. It is necessary to be penetrated by the principle, sufficiently obvious but often forgotten, that a document only contains the ideas of the man who wrote it, and to make it a rule to begin by understanding the text by itself, before asking what can be extracted from it for the purposes of history. We thus arrive