Here is a circular choir, surrounded by two rows of columns, irregularly spaced, with circular chapels outside, which seems to have been more or less what the architect of Chartres, for the Virgin’s purposes, had set his heart on obtaining.  Closely following the scheme of Saint-Martin-des-Champs came the scheme of the Abbey Church at Vezelay, built about 1160-80.  Here the vaulting sprang directly from the last arch of the choir, as is shown on the plan, and bearing first on the light columns of the choir, which were evenly spaced, then fell on a row of heavier columns outside, which were also evenly spaced, and came to rest at last on massive piers, between which were five circular chapels.  The plan shows at a glance that this arrangement stretched the second row of columns far apart, and that a church much larger than Vezelay would need to space them so much farther apart that the arch uniting them would have to rise indefinitely; while, if beyond this, another aisle were added outside, the piers finally would require impossible vaulting.

[Illustration with caption:  Vezelay]

The problem stood thus when the great cathedrals were undertaken, and the architect of Paris boldly grappled with the double aisle on a scale requiring a new scheme.  Here, in spite of the most virtuous resolutions not to be technical, we must attempt a technicality, because without it, one of the most interesting eccentricities of Chartres would be lost.  Once more, Viollet-le-Duc:—­

As the architect did not want to give the interior bays of the apse spaces between the columns (AA) less than that of the parallel bays (BB), it followed that the first radiating bay gave a first space (LMGH) which was difficult to vault, and a second space (HGEF) which was impossible; for how establish an arch from F to E?  Even if round, its key would have risen much higher than the key of the pointed archivolt LM.  As the second radiating bay opened out still wider, the difficulty was increased.  The builder therefore inserted the two intermediate pillars O and P between the columns of the second aisle (H, G, and I); which he supported, in the outside wall of the church, by one corresponding pier (Q) in the first bay of the apse, and by two similar piers (R and S) in the second bay.

[Illustration with caption:  Notre dame de Paris]

“There is no need to point out,” continued Viollet-le-Duc, as though he much suspected that there might be need of pointing out, “what skill this system showed and how much the art of architecture had already been developed in the Ile de France toward the end of the twelfth century; to what an extent the unity of arrangement and style preoccupied the artists of that province.”

In fact, the arrangement seems mathematically and technically perfect.  At all events, we know too little to criticize it.  Yet one would much like to be told why it was not repeated by any other architect or in any other church.  Apparently the Parisians themselves were not quite satisfied with it, since they altered it a hundred years later, in 1296, in order to build out chapels between the piers.  As the architects of each new cathedral had, in the interval, insisted on apsidal chapels, one may venture to guess that the Paris scheme hampered the services.