colour of each stone may be "true" that of the mosaic is false. Dezobry's Rome au siècle d'Auguste, Augustin Thierry's Récits mérovingiens, and other "pictures" produced at the same epoch were constructed on the same principle, and are subject to the same drawbacks as the historical novels properly so-called.
We may summarise what precedes by saying that, up to about 1850, history continued to be, both for historians and the public, a branch of literature. An excellent proof of this lies in the fact that up till then historians were accustomed to publish new editions of their works, at intervals of several years, without making any change in them, and that the public tolerated the practice. Now every scientific work needs to be continually recast, revised, brought up to date. Scientific workers do not claim to give their works an immutable form, they do not expect to be read by posterity or to achieve personal immortality; it is enough for them if the results of their researches, corrected, it may be, and possibly transformed by subsequent researches, should be incorporated in the fund of knowledge which forms the scientific heritage of mankind. No one reads Newton or Lavoisier; it is enough for their glory that their labours should have contributed to the production of works by which their own have been superseded, and which will be, sooner or later, superseded in their turn. It is only works of art that enjoy perpetual youth. And the public is well aware of the fact; no one would ever think of studying natural history in Buffon, whatever his opinion might be of the merits of this stylist. But the same public is quite ready to study history in