Skip to content

The Construction of Gothic Cathedrals

Post date:

In 1995 with too much time on my hands and living in King’s Cross, I borrowed the book of the title from Camden Library around the corner. I remember being amazed how someone had used observations of the buildings plus the suppositions of others such as Eugène-Emmanuel Viollet-le-Duc and Auguste Choisy, to deduce how Gothic cathedrals, and the vaults in particular, had been constructed. Gothic cathedrals were the most complex stone masonry structures the world had ever seen and, in all the centuries since, nobody has understood stone masonry construction as well and, because there’s no need today, probably ever will. It’s true we can make some plausible deductions about how The Pyramids were constructed but Gothic cathedrals were far more complex. We can only make informed guesses. Master masons didn’t record or publicize their methods, or give away any secrets. Below is a long quote from page 3 that’s better than any introduction of mine.

Like any scholar, the author weighs the assumptions of others who have approached the subject, before offering his own conclusions. First published in 1961, it’s written in 1960s academic language with meaningful commas and long, information-dense sentences that, though not inaccessible, don’t make for easy reading. I did have to dust off my vocabulary of medieval stone masonry construction terms and I did have to make sure I understood the meaning of each sentence before proceeding but this is my fault and not the author’s.

Because a given groin voussoir must accommodate itself to the conditions of both vaults, its outline, seen from above, is V-shaped; but the upper and lower surfaces are non-parallel, on the one hand, and broken other than being in the same plane, on the other. In addition, every groin voussoir has two intradoses, one for each of the intersection vault surfaces, and these intradoses of course are curving. Each grown voussoir is therefore a ten-sided stone, no two surfaces of which are parallel, and four of which are curving


The author is referring to voussoirs in vaults such as those in the image on the left, below, but his description only comes alive when the image shows the voussoirs, as in the image on the right. This is still the introductory chapter so the author doesn’t go so far as to mention the shaping of the groin vault keystones. This is probably something only the master mason or possibly also some trusteed team members knew.

The safety margin in those voussoirs must have been known because their elaborately carved ornament is necessarily achieved by subtraction of volume and was probably not performed by the mason who laid out their geometry and shaped or oversaw the shaping of the structurally portions structurally necessary. Ornament is rarely structural but, going by the progressive incline of the upper surfaces, ornamented protrusions such as those in the drawing on the right below (by Viollet-le-Duc) structural (as well as visual) continuity with the ribs and this was then transferred to capitals and the column or pilaster shafts. The uppermost stone in the image B below left shows the structural principle while the image on the right shows this ornamented transition.

As the sub-title says, this book is A Study of Mediaeval Vault Erection and so it mentions flying buttresses mostly with respect to their role in achieving those two Gothic cathedral driving aspirations of maximum height and maximum light because, with stone masonry construction, achieving maximum light means shifting the wall supports to outside the plane of the windows. As a means of understanding the development of construction techniques, the author looks at the aesthetic development of vaults in order arrive at an understanding of at what construction developments made them possible. The aesthetic developments of gothic vaulting are pretty much all we ever teach about them when we do, and this reveals our limited view of technology, innovation and human ingenuity.

The ribbed barrel vault may have been an aesthetic improvement on the barrel vault because of the perspective effect when seen from below, but it also meant that construction could take place in stages and allow the nave to be used – and hence donations received – while the remainder of the nave was still being constructed. The ribbed barrely vault may well be construction, aesthetic and funding innovation in equal parts. We would do well to remember that technical innovation usually means lower costs for someone, and that the more aesthetically progressive some technological innovation is touted as being, the more suspicious we should be. It’s the same with the ribbed groin vault. Beautiful they may be but – and this is contentious – once the groin ribs have been constructed using centering, that centering can be removed and the ribs themselves used as centring for corbelled stones above. Another example is the semicicular domed groin vault that produced less horizontal thrust at its supports, and therefore required less massive abutments. I would go so far as to say that aesthetics was and continues to be more the justification for technical advances rather than it is their driver.

Some more things I learned:

  • I can understand how there might not have been much demand for gothic cathedrals because they were, after all, huge commitments of time and resources for a several generations over century or two. Nevertheless, master masons were in short supply and wanted to keep it that way by not giving away any trade secrets. Gothic cathedrals are the sum of knowledge of “not just one particular region but of the entire era”.
  • I’d always thought large horizontal tie beams between rafters would act in tension to prevent the roof from spreading but tie rods could have done the same job. The large tie beams of gothic cathedral roofs were detailed so they would also act in compression, making the roof a rigid frame distributing wind forces to the buttresses on the other side. They’re there to stop the vaults from collapsing but transferring the wind loads to them means they can work a bit harder in the way they were intended to. Spreading the wind load was important when the pitch of these roofs was anything from 45° to as much as 65°.
  • The forces in the walls of gothic cathedrals of course changed in magnitude as construction progressed but they also changed in direction when construction of the vaults began. This is obvious, but I had to be told. It meant that builders had an awareness of the magnitude and direction of forces acting in any particular building elements at any given time, and adjusted the amount of temporary support (or counterweight) accordingly. This must have been hard-earned knowledge as mistakes were surely made.
  • There were plenty of complex construction problems that needed to be solved but there were also simple ones that were solved in simple ways. This drawing onj the left below, shows how the void behind the vault was simply backfilled so that forces pushing it outwards were directly transferred to the buttresses. The drawing on the right is my favorite in the entire book. It shows you how to create a vault without formwork – provided of course that your stonecutting (for a vault non-circular in section) is accurate.
  • The various galleries and spiral stairs [called vices, I learn] were important for conveying materials and formwork to the upper levels during construction, and also for enabling routine post-construction tasks such as maintenance and repairs without the need for large temporary formwork.
  • In the same vein, the roof structure would have had to be used to place large keystones to complete the vaults.

Most of this knowledge is either lost or was never recorded. We have Viollet-le-Duc, Auguste Choisy and John Fitchen to thank for being curious enough to think about the processes and sequences by which gothic cathedrals and especially their vaults were constructed. We know that they were. We just don’t know how.

• • • 

It’s said the 2019 fire at Notre-Dame de Paris cathedral was brought under control before more serious damage occurred but what this means is that the stone vaults protected the interior from the fire even though the roof and Eugene Viollet-le-Duc’s 93-meter-high spire were completely destroyed. Only a few portions of the vaults collapsed but the one below Viollet-le-Duc’s spire suffered the most.

One reason for the limited damage was because the tactical firefighting was mostly performed from inside using low-pressure hoses despite the increased danger to firefighters from falling timbers, collapsing vaults, dripping molten lead and such. President Macron’s 2019 promise to have Notre-Dame de Paris reconstructed within five years was at the time a wish rather than a statement showing any understanding of the challenges involved.

The construction of Notre-Dame de Paris began in 1163 and was largely completed by either 1260 or 1345 depending on your sources, although alterations and additions continued for centuries after. One logistical problem was how to authentically replicate in five years the construction of something that originally took maybe one and a half centuries to build. It might be possible to source wood from 2,000 authentic trees from still-extant forests but was this wood going to be fashioned into timbers before it had been seasoned for a century? Building Gothic cathedrals required time and forward planning. Another question is whether the original lead roof was really going to be replaced as-was? Maybe the original builders would have preferred to use titanium had they known about it and had no problem with dredging the ocean floor for titanium sands? The biggest unknown was how the vaults were constructed as there was more to it than simply getting some shaped rocks into place and putting a roof on top. The forces within the walls and buttresses were continually changing in magnitude and in direction as construction progressed and it was the roof that completed the building and equalized those forces prior to adding the keystones to the vaults. Now that the rebuilding is mostly complete, there can’t have been be any need for contemporary rebuilders, repairers and restorers to scrimp on formwork for surface finishing when, as long as the floor is protected, a 45-metre lift boom (a.k.a. cherry picker) can reach well over the 32-metre high vaults. The problem is that keystones can’t be placed from below.

It’s fairly certain that temporary loads from centering were applied to external walls to maintain maximum compression in them during construction as well as for enabling a certain amount of reuse and advance repositioning. When it comes to reconstructing the roof of Notre Dame de Paris, lifting completed trusses into place might not be the wisest thing to do. I don’t know. I hope somebody did.

John Fitchen, the author of The Construction of Gothic Cathedrals, died in 1989. I hope all the relevant people furiously studied The Construction of Gothic Cathedrals since 2019 and attempted to attain the author’s level of admiration for, and understanding of how these buildings were constructed.

• • •

Needless to say, I learned much about the construction of vaults in Gothic cathedrals but one nagging question is why were vaults necessary at all if there was a perfectly fine weatherproof roof above them? The small illustration on the title page shows that the forces tending to spread the vaults are transferred to the buttresses but, as they are, they also keep the high walls spaced and vertical even though the large roof tie beams do the same job. Earlier, I mentioned how some sort of timber structure above the vaults is necessary for the keystones to be placed and so complete the vaults but, if this were the only reason, the roof would be little more than waterproofed residual scaffolding. I also mentioned how the heavy tie beams transfer wind loads (and the forces withstanding them) from one side of the building to the other. This is something the vaults can’t do, and they also can’t help the roof to do. In a building in which all parts have multiple functions either at the same time or over time, I don’t understand why the vaults need the roof or at least some significant cross-timbers in order to be constructed, but the roof doesn’t need the vaults in order to be constructed. Either I’m not understanding everything or I’m overthinking it. Maximum height and maximum light aren’t prerequisites for worship but I can understand how they might help set the mood. Having the verticals of the columns transform into ribs vanishing to a point would have added symbolic value but similar effect could have been achieved with some kind of timber construction. If this had ever been an option, it suggests one should never trust an architect or builder to recommend the cheapest or easiest solution. Whichever way the question is asked, questioning why the vaults are there seems sacrilegious. It’s good to know that it’s knowledge and skill and not faith that keeps all that rock up in the air thirty or so meters above worshippers’ heads.

• • •


  • Superb stuff. I can never understand how drawings…and models…don’t survive. Information had to be communicated from craftsmen to craftsmen and labourers….over time! Where’s the stuff?!

    • says:

      Yes, this is something I don’t understand either. Building had come a long way from simply drawing lines in the sand using a stick. I can’t begin to think how I’d mark out a piece of rock for carving into a 16-sided voussoir with two curved surfaces and no parallel ones. Perhaps some form of wood or maybe skin template was used? Each vault would have had to have four sets of identical voussoirs and that set would need to be repeated for any number of other vaults. Such templates likely perished, or more likely, were destroyed once used. Some kind of model or sketch would have communicated the desired result to workmen and perhaps clients but these wouldn’t have been any use in understanding the structural forces involved. Teaching shows me time and time again that it’s easier to make a model to represent a structure than it is to represent the forces acting in that structure. I’ve accepted that being able to vizualise structural forces is a special skill and it’s probably unreasonable to expect all undergraduate students to have it or learn it.