[168] Primitive Greece has been studied in the Homeric poems. Mediæval private life has been reconstructed principally from the _chansons de gestes_. (See C. V. Langlois, _Les Traditions sur l'histoire de la société française au moyen âge d'après les sources littéraires_, in the _Revue historique_, March-April, 1897.) [169] Most historians refrain from rejecting a legend till its falsity has been proved, and if by chance no document has been preserved to contradict it, they adopt it provisionally. This is how the first five centuries of Rome are still dealt with. This method, unfortunately still too general, helps to prevent history from being established as a science. [170] For the logical justification of this principle in history see C. Seignobos, _Revue Philosophique_, July-August 1887. Complete scientific certitude is only produced by an agreement between observations made on different _methods_; it is to be found at the junction of two different paths of research. [171] This case is studied and a good example given by Bernheim, _Lehrbuch_, p. 421. [172] It is hardly necessary to enter a caution against the childish method of counting the documents on each side of a question and deciding by the majority. The statement of a single author who was acquainted with a fact is evidently worth more than a hundred statements made by persons who knew nothing about it. The rule has been formulated long ago: _Ne numerentur, sed ponderentur_. [173] Cf. _supra_, p. 94. [174] It is hardly possible to study here the special difficulties which arise in the application of these principles, as when the author, wishing to conceal his indebtedness, has introduced deviations in order to put his readers off the scent, or when the author has combined statements taken from different documents. [175] Here we merely indicate the principle of the method of confirmation; its applications would require a very lengthy study. [176] Père de Smedt has devoted to this question a part of his _Principes de la critique histoire_ (Paris, 1887, 12mo). [177] The solution of the question is different in the case of the sciences of direct observation, especially the biological sciences. Science knows nothing of the possible and the impossible; it only recognises facts which have been correctly or incorrectly observed: facts which had been declared impossible, as the existence of aerolites, have been discovered to be genuine. The very notion of a miracle is metaphysical; it implies a conception of the universe as a whole which transcends the limits of observation. (See Wallace, "Miracles and Modern Spiritualism.") [178] See above, p. 194. [179] In the experimental sciences an hypothesis is a form of question accompanied by a provisional answer.