On the other hand, men whose information is all that could be desired, whose monographs intended for specialists are full of merit, sometimes show themselves capable, when they write for the public, of grave offences against scientific method. The Germans are habitual offenders: consider Mommsen, Droysen, Curtius, and Lamprecht. The reason is that these authors, when they address the public, wish to produce an effect upon it. Their desire to make a strong impression leads them to a certain relaxation of scientific rigour, and to the old rejected habits of ancient historiography. These men, scrupulous and minute as they are when they are engaged in establishing details, abandon themselves, in their exposition of general questions, to their natural impulses, like the common run of men. They take sides, they censure, they extol; they colour, they embellish; they allow themselves to be influenced by personal, patriotic, moral, or metaphysical considerations. And, over and above all this, they apply themselves, with their several degrees of talent, to the task of producing works of art; in this endeavour those who have no talent make themselves ridiculous, and the talent of those who have any is spoilt by their preoccupation with the effect they wish to produce. Not, let it be well understood, that "form" is of no importance, or that, provided he makes himself intelligible, the historian has a right to employ incorrect, vulgar, slovenly, or clumsy language. A contempt for rhetoric, for paste diamonds and paper flowers, does not exclude a taste for a pure and strong, a terse and pregnant style. Fustel de Coulanges was a good writer, although throughout his life he recommended and practised the avoidance of metaphor. On the contrary we see no harm in repeating[230] that the historian, considering the extreme complexity of the phenomena he undertakes to describe, is under an obligation not to write badly. But he should write _consistently_ well, and never bedeck himself with finery. CONCLUSION I. History is only the utilisation of documents. But it is a matter of chance whether documents are preserved or lost. Hence the predominant part played by chance in the formation of history.