[140] A parallel difficulty occurs in the interpretation of illustrative monuments; the representations are not always to be taken literally. In the Behistun monument Darius tramples the vanquished chiefs under foot: this is a metaphor. Mediæval miniatures show us persons lying in bed with crowns on their heads: this is to symbolise their royal rank; the painter did not mean that they wore their crowns to sleep in. [141] A. Boeckh, in the _Encyclopædie und Methodologie der philologischen Wissenschaften_, second edition (1886), has given a theory of _hermeneutic_ to which Bernheim has been content to refer. [142] The method of extracting information on external facts from a writer's conceptions forms part of the theory of constructive reasoning. _See_ book iii. [143] For example, Père de Smedt, Tardif, Droysen, and even Bernheim. [144] Descartes, who came at a time when history still consisted in the reproduction of pre-existing narratives, did not see how to apply methodical doubt to the subject; he therefore refused to allow it a place among the sciences. [145] Fustel de Coulanges himself did not rise above this kind of timidity. With reference to a speech attributed to Clovis by Gregory of Tours, he says: "Doubtless we are unable to affirm that these words were ever pronounced. But, all the same, we ought not to affirm, in contradiction to Gregory of Tours, that they were not.... The wisest course is to accept Gregory's text" (_Monarchie franque_, p. 66). The wisest, or rather the only scientific course, is to admit that we know nothing about the words of Clovis, for Gregory himself had no knowledge of them. [146] Quite recently, E. Meyer, one of the most critically expert historians of antiquity, has in his work, _Die Entstehung des Judenthums_ (Halle, 1896, 8vo), revived this strange juridical argument in favour of the narrative of Nehemiah. M. Bouché-Leclercq, in a remarkable study on "The Reign of Seleucus II. (Callinicus) and Historical Criticism" (_Revue des Universités du Midi_, April-June 1897), seems, by way of reaction against the hypercriticism of Niebuhr and Droysen, to incline towards an analogous theory: "Historical criticism, if it is not to degenerate into agnosticism--which would be suicidal--or into individual caprice, must place a certain amount of trust in testimony which it cannot verify, as long as it is not flatly contradicted by other testimony of equal value." M. Bouché-Leclercq is right as against the historian who, "after having discredited all his witnesses, claims to put himself in their place, and sees with their eyes something quite different from what they themselves saw." But when the "testimony" is insufficient to give us the scientific knowledge of a fact, the only correct attitude is "agnosticism," that is, a confession of ignorance; we have no right to shirk this confession because chance has permitted the destruction of the documents which might have contradicted the testimony.