Here we must take leave of Normandy; a small place, but one which, like Attica or Tuscany, has said a great deal to the world, and even goes on saying things—not often in the famous genre ennuyeux—to this day; for Gustave Flaubert’s style is singularly like that of the Tour Saint-Romain and the Abbaye-aux-Hommes. Going up the Seine one might read a few pages of his letters, or of “Madame de Bovary,” to see how an old art transmutes itself into a new one, without changing its methods. Some critics have thought that at times Flaubert was mesquin like the Norman tower, but these are, as the French say, the defects of his qualities; we can pass over them, and let our eyes rest on the simplicity of the Norman fleche which pierces the line of our horizon.
The last of Norman art is seen at Mantes, where there is a little church of Gassicourt that marks the farthest reach of the style. In arms as in architecture, Mantes barred the path of Norman conquest; William the Conqueror met his death here in 1087. Geographically Mantes is in the Ile de France, less than forty miles from Paris. Architecturally, it is Paris itself; while, forty miles to the southward, is Chartres, an independent or only feudally dependent country. No matter how hurried the architectural tourist may be, the boundary-line of the Ile de France is not to be crossed without stopping. If he came down from the north or east, he would have equally to stop,—either at Beauvais, or at Laon, or Noyon, or Soissons,—because there is an architectural douane to pass, and one’s architectural baggage must be opened. Neither Notre Dame de Paris nor Notre Dame de Chartres is quite intelligible unless one has first seen Notre Dame de Mantes, and studied it in the sacred sources of M. Viollet-le-Duc.
Notre Dame de Mantes is a sister to the Cathedral of Paris, “built at the same time, perhaps by the same architect, and reproducing its general dispositions, its mode of structure, and some of its details”; but the Cathedral of Paris has been greatly altered, so that its original arrangement is quite changed, while the church at Mantes remains practically as it was, when both were new, about the year 1200. As nearly as the dates can be guessed, the cathedral was finished, up to its vaulting, in 1170, and was soon afterwards imitated on a smaller scale at Mantes. The scheme seems to have been unsatisfactory because of defects in the lighting, for the whole system of fenestration had been changed at Paris before 1230, naturally at great cost, since the alterations, according to Viollet-le-Duc (articles “Cathedral” and “Rose,” and allusions “Triforium"), left little except the ground-plan unchanged. To understand the Paris design of 1160-70, which was a long advance from the older plans, one must come to Mantes; and, reflecting that the great triumph of Chartres was its fenestration, which must have been designed immediately after 1195, one can understand how, in this triangle of churches only forty or fifty miles apart, the architects, watching each other’s experiments, were influenced, almost from day to day, by the failures or successes which they saw The fenestration which the Paris architect planned in 1160-70, and repeated at Mantes, 1190-1200, was wholly abandoned, and a new system introduced, immediately after the success of Chartres in 1210.