we content ourselves, in practice, with an abridged method.
All words are not equally subject to variations of meaning; most of them keep a fairly uniform meaning in all authors and in all periods. We may therefore be satisfied to study specially those expressions which, from their nature, are liable to take different meanings: first, ready-made expressions which, being fixed, do not follow the evolution of the words of which they are composed; secondly, and chiefly, words denoting things which are in their nature subject to evolution; classes of men (miles, colonus, servus); institutions (conventus, justitia, judex); usages (alleu, bénéfice, élection); feelings, common objects. In the case of all words of such classes it would be imprudent to assume a fixed meaning; it is an absolutely necessary precaution to ascertain what is the sense in which they are used in the text to be interpreted. "These studies of words," said Fustel de Coulanges, "have a great importance in historical science. A badly interpreted term may be the source of serious error." And, in fact, simply by a methodical application of interpretative criticism to a hundred words or so, he succeeded in revolutionising the study of the Merovingian epoch.
IV. When we have analysed the document and determined the literal meaning of its phrases, we cannot even yet be sure that we have reached the real thoughts of the author. It is possible that he may have used some expressions in an oblique sense; there are several kinds of cases where this occurs: allegory and symbolism, jests and hoaxes, allusion and implication, even the ordinary figures of speech, metaphor, hyperbole, litotes. In all these cases it