In his interesting book on this subject,[4] M. Stapfer has not taken into account changes in civilization. He has thought only of chance to explain the loss of so many ancient books. Chance is a mask, and it is precisely the duty of the historian to lift this mask, or to tear it. Between the sixth century and our own day, there has been one further partial modification in civilization--in the fifteenth century. About that time, the old literature began to lose its hold upon the public. The novels, the miracles, the tales seemed suddenly to have aged. They were no longer copied or recited. They were seldom printed, a single manuscript having preserved for us _Aucassin et Nicolette_, which is something like the _Daphnis and Chloe_ of the Middle Ages. Accidents frighten the poet--and even the critic, who is colder, whose logic is more rigorous--the moment the suggestion is made of separating the purely historical idea of literary survival from the sentimental idea of justice. Till now--and I allude once more to the conservative rôle of modern civilization--the printing-press has protected writers against destruction; but the serious rôle of printing affects as yet only four centuries. This distant invention will appear some day as if contemporary at once with Rabelais and with Victor Hugo. When a time equal to that which separates us from the birth of Æschylus--two thousand three hundred and seventy-five years, let us say--shall have elapsed between us and a given moment of the future, what influence will printing have had on the preservation of books? Perhaps none. Everything not worth the trouble of reprinting--that is to say everything, with the exception of a few fortunate fragments--will have disappeared, and the more rapidly that the material substance of books has become more precarious. Even the discovery of a durable paper would not give absolute assurance of survival, because of the temptation to employ this excessively strong paper for a thousand other purposes. Thus the value of the parchment has often led to the sacrifice of a manuscript, just as gold articles go necessarily to the smelting-pot once the style has changed. The best material for the preservation of books would be something unchanging, but fragile, slightly brittle, so that it would be good for nothing outside its binding. Would not such a discovery be a curse?