[157] It will be seen why we have not separately defined and studied "first-hand documents." The question has not been raised in the proper manner in historical practice. The distinction ought to apply to _statements_, not to documents. It is not the document which comes to us at first, second, or third hand; it is the statement. What is called a "first-hand document" is nearly always composed in part of second-hand statements about facts of which the author had no personal knowledge. The name "second-hand document" is given to those which, like the work of Livy, contain nothing first-hand; but the distinction is too crude to serve as a guide in the critical examination of statements. [158] There is much less modification where the oral tradition assumes a regular or striking form, as is the case with verses, maxims, proverbs. [159] Sometimes the _form_ of the phrase tells its own tale, when, in the midst of a detailed narrative, obviously of legendary origin, we come across a curt, dry entry in annalistic style, obviously copied from a written document. That is what we find in Livy (see Nitzsch, _Die römische Annalistik_, Leipzig, 1873, 8vo), and in Gregory of Tours (see Loebell, _Gregor von Tours_, Leipzig, 1868, 8vo). [160] The events which strike the popular imagination and are transmitted by legend are not generally those which seem to us the most important. The heroes of the _chansons de gestes_ are hardly known historically. The Breton epic songs relate, not to the great historical events, as Villemarqué's collection led people to believe, but to obscure local episodes. The same holds of the Scandinavian sagas; for the most part they relate to quarrels among the villagers of Iceland or the Orkneys. [161] The theory of legend is one of the most advanced parts of criticism. Bernheim (in his _Lehrbuch_, pp. 380-90) gives a good summary and a bibliography of it. [162] "History of Greece," vols. i. and ii. Compare Renan, _Histoire du peuple d'Israël_, vol. i. (Paris, 1887, 8vo), Introduction. [163] And yet Niebuhr made use of the Roman legends to construct a theory, which it was afterwards necessary to demolish, of the struggle between the patricians and the plebeians; and Curtius, twenty years after Grote, looked for historical facts in the Greek legends. [164] See _supra_, pp. 93 _sqq._ [165] Cf. _supra_, p. 166. [166] It is often said, "The author would not have dared to write this if it had not been true." This argument does not apply to societies in a low state of civilisation. Louis VIII. dared to write that John Lackland had been condemned by the verdict of his peers. [167] See above, p. 153. Similarly, the particular facts which compose the history of forms (palæography, linguistic science) are directly established by the analysis of the document.