The Flying Buttress
The stone arches propping a cathedral from the outside are the reason the wall inside could become glass.

Walk round to the east end of Notre-Dame de Paris and the building briefly stops looking like a building. A ring of stone arches springs out of the upper wall, clears the roof of the chapels below and lands on free-standing piers set out among the trees, so that the choir appears to be held up by something standing several metres away from it. The arches are called flyers, the piers they land on are the buttresses proper, and between them they do the one job the wall was relieved of. They catch a push that a stone ceiling makes sideways.
A vault does not only press down. Any arch carried across a span in masonry delivers part of its load outward where it springs, and a vault is a field of arches, so that outward push runs along the whole head of the wall for the length of the building. Romanesque builders answered it with mass, giving their churches walls thick enough to swallow the thrust before it could tip them and therefore windows small enough not to weaken those walls, which is why a Romanesque nave is dim. The Gothic answer began with two changes made inside. A pointed arch stands more steeply than a round one and throws less of its load sideways, and a rib vault gathers what remains onto a handful of points rather than spreading it along a continuous wall-head. The thrust was now concentrated, predictable and arriving in known places.
Concentrated thrust can be intercepted. The flyer meets it just under the roofline, carries it down a shallow arc across the aisle and hands it to a pier standing clear of the building, which walks it to the ground. Between one flyer and the next the wall has almost nothing left to hold up, and the builders spent the difference on window. Flying buttresses begin to appear in the 1170s, a generation after Abbot Suger rebuilt the ambulatory of Saint-Denis between 1140 and 1144 and showed France what a church looked like when its walls had stopped carrying the roof. Suger wanted the light for theological reasons. The flyer is how the twelfth century paid for it.
For a long time nobody could say whether the flyers at Notre-Dame belonged to the plan or to a rescue. They are single-span and unusually long, and to several scholars they looked like a later fix bolted on once the walls began to lean outward. Stephen Murray, at Columbia, argued instead that they had been intended from the start of work around 1160. In 2010 Andrew Tallon, an art historian at Vassar, scanned the whole cathedral with a laser, gathering over a billion measured points. The nave walls stood very nearly plumb, and walls that have leaned and then been caught do not come back true. The flyers were original, and the question was settled by a machine reading what the masons had left in the stone.
Five minutes standing inside the system rather than outside it. Watch how they read the wall upward in three storeys, arcade, triforium, clerestory, and notice the point where the pier that carries the vault is pointed out beneath its casing of thin decorative shafts. The flyer outside is the other half of that pier.
The pinnacles took much longer to explain. They are the small stone spires planted on top of each buttress pier, and for centuries they read as ornament with a stonemason’s bill attached. In 1968 Robert Mark, an engineer at Princeton, began cutting cathedral cross-sections out of transparent epoxy and squeezing them between polarising filters, a technique called photoelasticity in which stressed plastic prints its own strain in bands of colour. Loading a model of Amiens for wind as well as for its own weight, Mark and Richard Alan Prentke found tension opening in the buttress uprights on the sheltered side of the building, which is precisely where jointed masonry is weakest, and found that the dead weight of the pinnacle standing above cancelled it. The masons had no theory of tensile stress and no way to picture one. They had wind, and they had the memory of buildings coming down, and they put the stone where it turned out to be needed.
The collapses were real enough, and recent enough, to concentrate the mind. Through the thirteenth century the cathedral chapters of northern France pushed each other upward, each commissioning a vault a little taller than its neighbour’s, and Beauvais took the contest at forty-eight metres. In 1284, twelve years after its choir was finished, part of that vault came down and brought flying buttresses with it. The choir went back up at the same height on twice as many flyers, and no one in Europe built higher again. Everything the flyer had bought, the height and the thinness and the coloured light, turned out to have a limit, and Beauvais is where the masons found it.