The St-Philibert in Tournus (between 1020 an 1050) shows an interesting solution. The nave isn't covered by one barrel vault, but by 5 crosswise placed adjacent barrel vaults.[br]By doing this the side walls of the church aren't pushed aside. The barrel vaults neutralize the horizontal forces of each other.

In Tournus the side walls didn't have to support the horizontal forces of the vaults. As a result windows could be placed higher an could be enlarged. Side aisles provided supplementary support for the high walls of the church.

Technicly the soluton is perfect. But the experiment of Tournus was never to be repeated. The only explanation for this is that esthetically it wasn't satisfying. The weight of the barrel vaults got concentrated unto the round piers by huge crosswise placed arcs. But these arcs devided the unity if the chuch into separated parts and the unity of the nave got lost. Both technique and esthetics influenced the evolution of romanesque architecture. Only these solutions that satisfied both aspects did survive and were repeated in the following projects.