Beauvais has always been a convenient example and something of a parable for structures professors. It illustrates the race for height that bishops and towns felt in the 13th century–its choir is the tallest by six meters–and it has also traditionally illustrated the limits of masonry in the gothic system. The choir was built from 1225 to 1272, but a large segment of the vaulting collapsed in 1284, effectively stopping work. The vaults were rebuilt with additional piers and with sexpartite vaulting instead of quadripartite (in other words, the single vaults along the sides of the roof were doubled up). And, at some point, iron tie rods were installed between the flying buttresses around the choir and the hemicycle, which survived the collapse almost intact. The tie rods that are there now were re-installed in the 1960s, a short time after the cathedral architect removed the originals as the buttresses began resonating in a windstorm. You can imagine the panic.
If you look closely, you can see that piers were added–trace the ribs that run through the vault keystones down, and in the lower story you can see that they don’t have full piers accompanying them. You can also see, if you look very closely, traces of the original pointed arches (they spanned two of the bays that are there now). And the new lower arches are lower than the surviving ones from the hemicycle, on the right side of the above picture. The center ribs, the two skinny vaults on either side of them, and the piers and colonnettes that descend from them were all added during the reconstruction of the vaults.
No Gothic builders tried to match Beauvais’ height or slenderness after the collapse, which supposedly demonstrates the empirical approach behind the era’s construction–build taller and taller until something falls down, and then build a bit shorter. That’s been shown to be largely untrue, though. There were plenty of socio-political and economic reasons for the retreat from truly giant cathedrals after the 12th century–the Hundred Years’ War and the Black Death among them. But Beauvais has also been something of a parlor game among historians trying to prove what, exactly, the mechanism of the collapse was. There are no surviving accounts of the collapse, so every theory has had to start from the fabric itself and try to find flaws that might have doomed the choir. Jacques Heyman proved in the 1960s that the structure was perfectly stable under dead loading, which naturally focused attention on wind as the culprit. As I understand it (and I’m happy to be corrected by the readership, of course) both he and Viollet-le-Duc blamed settlement in the towers for a change in geometry and thus the collapse–Heyman suggests geotechnical problems while Viollet-le-Duc thought it could be due to compression of lime mortar over time.
Robert Mark’s photoelastic analysis in the 1970s, though, suggested that wind loading on the slender piers would have actually induced tension on the windward piers–in other words, a strong enough wind would have actually started to pry the upper stories of the choir off of their supports. Over twelve years, Mark and and Maury Wolf argued, wind forces rocked the upper story back and forth, eventually weakening the piers and buttresses through repetitive dynamic stress. This would have weakened the mortar enough to allow a shear failure as imagined by Viollet-le-Duc or Heyman, albeit with a more straightforward cause. Mario Salvadori, among others, doubted Mark’s analysis, noting that the methodology used by Mark in this and in many other studies–loading two-dimensional acrylic and photographing the results with a polarized lens to reveal contours of stress–could show only the conditions of a monolithic structure, not one composed of hundreds of weakly bonded stones.
Reading all of this (on the train to Beauvais, which is a gentle 1-1/2 hour milk run), I found Mark’s rationale unconvincing. He argues that the repairs–adding additional piers and buttresses and changing the vaults from quadripartite to sexpartite–were common sense responses to the uplift problems. But they would also have been understandable responses if there had been a dead load issue, in other words, if the builders had thought simply that the vaults had been too heavy for the widely-spaced, slender supports that were there. After all, it would be another 700 years before Heyman would come along to prove that gravity wasn’t the problem.
Steven Murray makes a convincing suggestion, pointing out that the intermediate piers were basically hinged above the side aisles–you can see in the section that the middle piers get quite narrow as they descend through the cathedral’s occupied spaces. This, to his analysis, was a fundamentally weak point of the section and a likely point of rotation and failure.
All of these theories, interestingly, focus on the section as the problem. And this is understandable given the tools available during the 1960s through 1990s. Mark’s photo elastic analysis was cutting edge, but it necessarily required dividing the structure up into planes that could be routed out of acrylic. Heyman’s and Murray’s analyses are also based, fundamentally, on a more or less planar analysis. That’s entirely understandable, since that’s the way Beauvais has been represented in the literature since it was built. Its section is iconic, rich and complex on its own. And that complexity is multiplied exponentially when the section is repeated and rotated into the three-dimensional network of vaults and buttresses that you actually see when you walk in and around it.
A forensic historian today (and boy, if that’s a real job, I want in) has a different set of tools that should enable a more complex three-dimensional analysis, and I suspect that the role of the flying buttresses’ dimensions in the choir’s longitudinal direction might seem far more important. In particular, theories that rely on the post-collapse fixes for evidence should be taking into account not only the re-vaulting of the choir, but also the iron bars that were installed to brace the buttresses in the longitudinal direction (and, around the hemicycle, in the circumferential direction). Admittedly, we don’t know when these were installed, but they’re certainly not ‘original,’ and the reconstruction after the collapse seems a likely potential date. Given the scare of the 1960s, it’s clear that the slender stone buttresses were worryingly flexible in a good wind without that bracing. And one can imagine one of them rotating out of plane, thus moving the buttresses out of the line of thrust from the vault above, with obvious results. The rebuilders must have been nervous about gravity and wind, and may have added the intermediate ribs (turning the vaults into sexpartite instead of quadripartite) to solve perceived shortcomings in the former, and iron rods to solve the latter. The eventual construction of the transept, in the 16th century, provided additional longitudinal stability and would have sheltered the thin buttresses from some amount of wind.
Beauvais’ transept has its own tragic story, of course–the tower over the crossing collapsed in 1573, after which the bishop and the good people of Beauvais threw in the towel. That collapse is pretty well established, causation-wise, since the lack of a nave meant that the crossing had drastically uneven lateral support at its base. The restoration of the crossing is particularly good–new piers are rendered in a smoothly curving stone, distinguishing them from the original, fully articulated piers of the original, so you can see what survived and what didn’t.
All of that aside, what’s there is spectacular, and the proportions are, as advertised, utterly sublime. The discussion about the cause of the collapse can go on forever but the real question, standing in the middle of this space, is how on earth Beauvais’ builders got it to stand at all.
Further reading (most, but not all, on JSTOR):
Jacques Heyman, “Beauvais Cathedral,” Transactions of the Newcomen Society, vol. 40, 1967-68, pp. 15-36.
Stephen Murray, “The Choir of the Church of St.-Pierre, Cathedral of Beauvais: A Study of Gothic Architectural Planning and Constructional Chronology in Its Historical Context.” The Art Bulletin, Vol. 62, No. 4 (Dec., 1980), pp. 533-551.
Maury I. Wolfe and Robert Mark, “The Collapse of the Vaults of Beauvais Cathedral in 1284,” Speculum, Vol. 51, No. 3 (Jul., 1976), pp. 462-476.