A cathedral looks like an act of faith and a pyramid looks like an act of power, but both are really the same thing: an argument about a rock, carried out at enormous scale, by people who mostly could not read.
We are not going to begin with how to cut stone or how to lay it. We begin by asking what stone actually is, and what it will and will not put up with. The answer is one sentence long. Hold that sentence and every stone building you have ever walked past stops being decoration and starts being a diagram — of forces, drawn in rock, by someone trying to keep a very heavy thing in the air.
Pick up a stone. It is heavy for its size, cold, and it does nothing. That is the whole of most people's relationship with stone, and it is why almost nobody notices that a stone has a character — a set of things it is brilliant at and a set of things it is hopeless at. Everything a mason does is arranged around that character.
Start with what the trade actually is, because the word does a lot of quiet work. Masonry is building by stacking: separate pieces of stone or brick, laid one on another, held in place by nothing but their own weight and the weight of everything above them. There are no nails in it, no welds, no bolts. A masonry building is a very large, very carefully arranged pile. Every rule in the rest of this course is a rule about what a pile is allowed to do — and the pile takes its orders from the material, so the material comes first.
Look at what a stone is made of. Not chemistry — just picture. Stone is a great crowd of hard grains and crystals, pressed and grown together until they interlock. In many stones those grains were laid down as sand or shell or mud on some ancient floor and then buried under the weight of everything that landed on top afterwards, for a very long time. In others they cooled out of molten rock. Either way, the grains hold on to each other by fitting together and by a modest chemical grip at their edges.
Now compare that with two other materials, because the comparison is the lesson.
A stone is, if you like, an old squeeze that never let go. Keep squeezing it and it is content. Ask it to hold itself together against a pull and you are asking it to do the one thing nothing inside it was ever arranged to do.
Here is the sentence the rest of this course is built on. It is not difficult and it is not new — every mason who ever lived knew it in their hands long before anyone wrote it down.
Two words, defined here and used for the rest of the course. When something is being pressed or squeezed, we say it is in compression. When it is being drawn or stretched apart, we say it is in tension. That is all those words mean: squeezed, and stretched.
Stone in compression is astonishing. A block of good granite the size of your fist — call it four inches square on the face — will take something over a hundred tonnes pressed straight down on it before it crushes. Stone in tension is pitiful by comparison: for most building stones you need somewhere around ten times as much force to crush a piece as to pull it apart, and for many stones nearer twenty times. It is not a small difference in degree. It is two different materials wearing the same coat.
Now — and this is the move that makes the rest of the course inevitable — notice that a thing does not have to be pulled from both ends to be in tension. Bending puts one side of anything into tension. Take a bar of anything and bend it downwards in the middle: the top side gets shorter, so it is being squeezed, and the bottom side gets longer, so it is being stretched. Bending is a squeeze and a stretch, back to back, inside one piece.
Which gives us the sentence that will run this whole craft:
Every clever thing masons have invented over five thousand years — the arch, the dome, the buttress, the thick wall, the way a mortar joint is made — is a way of arranging a load so that no stone anywhere ends up being stretched. That is the entire game. From here on we are just watching people play it.
Do not take this on trust. There is a demonstration you can do in a yard with things you already own, and once you have done it you will never again wonder why stone buildings look the way they do.
That crack starting on the underside is the whole course in one event. The slab did not fail because it was weak. It failed because you found the one direction it cannot work in, and you found it by bending.
Follow the fact into its first consequence. Lay a flat stone across the top of two uprights and you have a lintel — the flat beam over a doorway or a window. It is the simplest way to bridge a gap and it is the first thing anybody tries.
But the middle of that lintel is unsupported and gravity is pulling it down. It bends, invisibly but really. Its top surface is squeezed, its underside is stretched, and the stretch is worst at the middle and worst on the very bottom face.
So a stone lintel has a hard ceiling on how far it can reach, and the ceiling is low. Make it longer and the bending gets worse quickly — much faster than the extra length would suggest. Make it thicker to fight back and you have added weight, which is more load, which is more bending. Stone is trying to solve its own problem with the very substance that is causing it.
This is why the great flat-stone buildings look the way they do. The uprights at Stonehenge carry lintels about ten feet long — call it three metres — and no longer. Greek temples are forests of columns not because the Greeks loved columns but because their marble beams could not reach further; put the columns further apart and the roof snaps. Greek builders sometimes hid iron bars inside those marble beams to help, which was a real and clever cheat and is exactly the same idea as modern reinforcement: put in something that can be pulled, precisely where the stretching happens.
Play with that until the pattern is obvious. Load the lintel and a red band grows along its underside until it opens. Load the arch and there is no red band anywhere, no matter how much you pile on. The arch simply presses harder. That is not a trick of the drawing. It is why the arch exists.
Here is the problem stated plainly. We must cross a gap. Crossing a gap with a straight beam means bending, and bending means stretching, and stretching is the one thing stone will not do. So: cross the gap without a straight beam.
The answer is to cut the stones as wedges and stand them in a curve, each leaning on the next. Each wedge is called a voussoir — an old French word that simply means one of the wedge-shaped stones of an arch. The top of the curve is the crown, and the wedge sitting at the crown is the keystone, which is the last one in.
Now think about what a single wedge feels. Its two flat faces are pressed by the wedges on either side. It is not hanging, it is not bending, it is not being pulled in any direction. It is being squeezed from both sides at once, and it responds by squeezing its neighbours in turn. The squeeze passes down the curve from the crown to the ground the way a whisper passes down a line of people.
That is the invention. Not the shape, which is only the visible part. The invention is the idea that you can convert an impossible job — resisting a bend — into an easy one, by changing the geometry so the job never arises. Nothing about the stone changed. The question changed.
It is worth sitting with how strange that is. Arches are made of pieces that individually can do nothing. A single voussoir on the ground is a heavy paperweight. A ring of them, standing, holds up a bridge that will still be there when everyone reading this is dust. The strength was never in the stones. It was in the arrangement.
An arch has a peculiar property that follows straight from Module 5: it does not work in stages. Half an arch is not half a bridge. Half an arch is a pile of wedges falling into a hole.
Until the last wedge is in, no voussoir has a neighbour on both sides, so no squeeze can run through. So every arch in history has been built on a temporary wooden former — a curved timber cradle called centring — that holds every stone in place until the ring is closed. When the ring is complete, the centring is knocked away, or "struck", and the arch takes its own weight for the first time.
Watch what happens at that moment, because it is the most honest few seconds in building. The arch settles a fraction. The joints close up tight. The wedges shuffle by a hair's breadth into the position where the squeeze runs cleanly, and then it stops moving and stands for a thousand years. An arch is not fixed rigid like a welded frame. It is a stack that finds its own comfortable line, and a small settlement is the sound of it finding it.
This also explains why the keystone has such a reputation. It is not magic and it is not stronger than the others — it is the same stone as its neighbours, cut to the same wedge. Its fame is that it is the piece that completes the chain. Until it is in, there is no arch at all; the moment it is in, there is. That is a good enough reason for the word to have walked out of the building trade and into ordinary speech, where a keystone is still the piece that makes the rest of it hold.
If the squeeze runs from the keystone down to the ground, we can ask a sharper question: where exactly inside the stone does it run? Draw that path and you have the single most useful picture in masonry. Masons and engineers call it the thrust line — the imaginary line traced by the push as it works its way down through the structure to the earth.
Two rules govern it, and both are consequences of the one fact.
First: the thrust line must stay inside the stone. If the line wanders out past a face of the masonry, there is nothing there to do the pushing, and the joint at that point simply opens and turns into a hinge. Enough hinges and the structure becomes a mechanism — a set of pieces free to rotate about each other — and it folds up. Masonry rarely fails by being crushed; the stone in a great cathedral pier — a free-standing block of masonry carrying a load down to the ground — is typically working at something like a tenth of what it would take to crush it, often far less. Masonry fails by shape. It fails when the line escapes.
Second: it is best if the line stays near the middle. If the push arrives at a joint dead centre, the whole face is squeezed evenly. Move it off centre and one edge of the joint is squeezed harder while the other relaxes — and if it goes far enough off centre, the far edge stops being squeezed at all and starts trying to open. Masons have a rule of thumb for this that needs no arithmetic: keep the thrust inside the middle third of the joint and the whole joint stays in compression. That is why it is called the middle third rule, and you can see it drawn in the demonstration in the next module.
Now the consequence that people miss, and it is the one that decides whether a building stands. Follow the squeeze down the curve of an arch and notice which way it is pointing when it gets to the bottom. At the crown the push is horizontal, wedge shoving wedge sideways across the top. By the time it reaches the ground it has swung round and is pointing mostly down — but not entirely down. It is still leaning outwards.
So an arch does not just sit on its supports. It shoves them apart. The place where the arch meets its support is called the springing, and at every springing there is a permanent outward push, day and night, for as long as the arch exists. That push is called the thrust.
This is not a flaw in the design. It is the price of the invention. You avoided tension inside the stone by curving the load — and curving the load is precisely what turned some of the downward weight into a sideways shove. You did not make the difficulty go away. You moved it to the feet, where you can do something about it.
That line is calculated, not decorated. It starts at the springing, over on the inner face of the pier, carrying the arch's outward shove and the arch's weight; and as it travels down it collects the weight of every course — every horizontal layer of stone — that it passes. More weight below means the line swings steeper, and steeper means it drifts sideways less before it lands. That is the entire secret of the next module.
If the thrust line must stay inside the masonry, there are exactly two ways to keep it there. Put more masonry where the line wants to go, or add weight above so the line runs steeper and wants to go somewhere nearer. That is all. Every heavy-looking thing on the outside of every stone building is one of those two moves.
And now you can answer a question nobody asked you. Why do arcades — long rows of arches — need nothing much between them, but need a great heavy pier or wall at each end? Because in the middle of a row, each arch's outward shove is met by its neighbour's shove coming the other way. They cancel. Only the two arches at the ends have nothing to lean against, so only the ends need an abutment — the stout mass of masonry that stands at the end of a run of arches and takes the shove that has nothing else to push against. That is why a Roman aqueduct is a line of slim piers with something distinctly heavier where it meets the hillside.
There is a third way to deal with the shove, and it is an admission that you have run out of stone. If two feet are being pushed apart, tie them together.
An iron tie rod strung across the springing of an arch takes the outward push directly, and iron is a material that is happy to be pulled — the exact opposite character to stone, which is why the two go together so well. Walk into almost any old Italian church and you will see iron rods stretched across the arches of the nave — the long main hall you walk up — some there since the day it was built and some added by a later generation who noticed a crack. There is no shame in either. It is the correct answer to a real force.
But there is a subtler answer, and it is the one that made Gothic architecture possible. Go back to your hanging chain. When you moved your fingers closer together, the curve got steeper and needed less pull to hold it. Turn that upside down: a steeper arch pushes outwards less.
That is the whole point of the pointed arch. It is not a style and it was not chosen because it looks aspiring. A pointed arch stands up steeper than a semicircle, so it delivers less sideways shove to its feet, so it needs less buttressing, so the walls can be thinner and the windows bigger and the whole thing taller. Every visual quality we associate with a Gothic cathedral — height, light, thinness — is downstream of one geometric fact about where the thrust line goes.
A pointed arch has a second gift. Two arches of different spans can be made to rise to the same height simply by changing how sharply they point, which means a rectangular bay can be vaulted without the ceiling going lopsided. A semicircular arch has no such freedom: its height is decided for it by its width. This is a genuine design constraint solved by a genuine geometric trick, and it is the sort of thing you can only notice once you are thinking in thrust lines rather than in styles.
Take the arch you now understand. Stand it up. Now spin it about its own vertical centre line, sweeping it all the way round, and let it leave a shell of stone behind as it goes. That is a dome. Not a metaphor — that is literally the geometry.
Everything you know carries over. Every slice through a dome from top to bottom is an arch, so the weight travels down curved paths in compression, so the stone is never bent, so the stone is happy. And the bill arrives in the same way as before: at the bottom edge, the dome pushes outwards all the way round the circle, in every direction at once. A ring of thrust, pointing out.
The Pantheon in Rome has stood with the largest unreinforced concrete dome on earth for about nineteen hundred years, and it is a catalogue of these ideas. The walls at the bottom are enormously thick to catch the ring of thrust. The material gets lighter as it goes up — the aggregate, meaning the broken stone mixed into the concrete, is heavy travertine and tuff low down and light volcanic pumice near the crown — so the shell weighs least exactly where weight would do most harm. The coffers, those sunken square panels, remove weight without removing much strength. And there is a hole in the top, which alarms people until they realise a dome does not need its crown to be closed: the rings near the top are squeezing themselves tight, so a neat round opening with a firm rim changes nothing important.
A dome invents one problem that no arch has, and it comes straight out of the spinning. Think about a horizontal ring of stone somewhere in the shell — one course, going right round.
Near the top, the shell is nearly flat and the rings are small. Each ring is being pressed inwards by everything around it, so the ring is squeezed. Stone is delighted.
But lower down, the shell is steep and spreading outwards. Each ring there is being forced to grow larger in circumference — the dome is trying to splay out at the bottom like a bell. A ring that is being made bigger is being stretched. That stretching round the circumference has a name worth knowing: hoop tension, the same force a barrel hoop resists when the barrel tries to burst. Below roughly fifty degrees down from the crown, a simple round dome carrying its own weight goes from squeezed rings to stretched rings.
And now the one fact takes its revenge. Stone cannot be stretched. So the rings crack — and because the stretch runs around the circumference, the cracks open up and down, like the segments of an orange. This is not a rare fault. It is normal. The Pantheon has them. Saint Peter's has them. Almost every large old masonry dome on earth is cracked into vertical segments.
Here is the calm part. A dome cracked into vertical slices has quietly turned itself back into something you already understand: a ring of separate arches leaning on each other. It can go on standing in that condition indefinitely, so long as something catches the outward shove at the bottom. And so the fix is the same fix as always, bent into a circle: put a ring around it.
There is an older way to roof a space in stone, and it looks similar enough that people confuse the two. It is worth separating them, because the difference is exactly the fact this course is built on.
To corbel is to lay each course of stone so it juts out a little further over the space than the course below, ring after ring, until the two sides meet overhead. The Treasury of Atreus at Mycenae is roofed that way. So is the burial chamber at Newgrange in Ireland, which has stayed dry for five thousand years. A snow house is built the same way, in a rising spiral, each block leaning on the last.
Now: a jutting stone is a small cantilever — a beam held at one end only, with the other end out in the air — and a cantilever bends. Which should be forbidden. So why does corbelling work at all? Because of what sits on the tail. Each projecting stone has the whole weight of the wall behind and above it pressing down on its inner end, holding it like a thumb on the end of a ruler. The stone is not free to tip and it is not free to bend much, because it is clamped by load. The corbel is not defying our one fact — it is being rescued from it by weight.
The tells, once you know them, are unmistakable. In an arch the joints run radially, pointing at the centre of the curve, and the stones are wedges. In corbelling the joints stay horizontal, the stones are ordinary flat blocks, and the profile is stepped rather than curved even when the steps are smoothed off afterwards. And corbelling needs a great mass of wall behind it to do the clamping, which is why corbelled chambers sit inside mounds and hillsides, while an arch can stand in the open air with almost nothing around it.
That contrast is the honest measure of the arch's cleverness. Corbelling gets you across a gap by burying the problem under sheer mass. The arch gets you across by changing the direction of the forces so the problem never occurs.
Nearly everyone believes mortar sticks stones together. It is the most common wrong idea in the whole craft, and our one fact demolishes it in a sentence: if mortar were a glue, its job would be to resist being pulled — and nothing in a stone wall is ever supposed to be pulled.
So what is the grey stuff for? Put two rough stones face to face and look at where they actually touch. Not everywhere. Three or four high points, maybe. All the weight of everything above must pass through those few knuckles, and a colossal load squeezed through a fingernail of contact will split the stone outwards from the point of contact — the same crack you would get by hitting it with a punch. Perfectly good stone, destroyed by touching in the wrong places.
Mortar is laid soft, the stone is bedded down into it, and the mortar squeezes out until the two faces are separated by a thin, even layer that fills every hollow. Now the load passes through the whole face instead of three knuckles. The mortar has not stuck anything to anything. It has made two imperfect surfaces behave like one perfect one.
Now derive the rules of good mortar without being told them:
The proof that mortar is not glue is that you can leave it out entirely. A dry stone wall has no mortar at all and stands for centuries. It works by choosing and turning each stone so it beds firmly on the ones below, by leaning both faces slightly inwards as it rises — a lean called the batter, roughly a couple of inches of lean for every foot of height — and by laying long through stones that reach right across from face to face. A through stone is not a rope: it does not pull the two faces together, it simply lies across both of them so that neither can go anywhere without the other. Nothing but gravity, friction and judgement. A dry stone waller is doing pure compression with the mortar deleted, and that is why it is the best possible apprenticeship for the eye.
If mortar's job is to be a soft, forgiving, even bed, then what it is made of matters enormously — and here is where a great deal of well-meant damage gets done to old buildings.
The old mortar is lime, and its story is a circle. You take limestone — which is a rock — and burn it in a kiln at around nine hundred degrees Celsius. The heat drives carbon dioxide out of it and leaves quicklime, a fierce, thirsty white material. Add water and it slakes: it hisses, swells, boils and turns into a soft putty. Mix that putty with sand and you have mortar. Then, laid in a wall, it slowly takes carbon dioxide back out of the air, month after month, and hardens back towards the limestone it started as. Burn a rock, water it, and let the air turn it into rock again. The circle closes in the wall.
Lime has two further properties that come from being soft and slightly porous, and both are exactly what an old wall wants. It lets water vapour pass, so damp that gets into a wall can leave again through the joints. And it is mildly self-healing: fine cracks fill themselves as free lime is carried into them by water and sets there. A lime wall is a slow, forgiving thing.
Be clear about what is doing the work in that paragraph, because it is not our one fact. Softness derives from the one fact — the joint must be the weaker part so that movement cracks the mortar and not the stone. Breathability does not. It is a second, independent fact, and it is about water rather than load: stone that cannot dry out is stone being taken apart by salt and frost. Both matter. Only the first one is a consequence of anything you have been told so far.
Modern cement mortar is a different animal. Portland cement, patented by Joseph Aspdin in Leeds in 1824, sets fast, sets very hard, and is nearly waterproof. For new work with modern materials it is excellent. Repointed into an old soft-stone wall — that is, raked out of the joints and replaced with the new material — it is a quiet disaster, for one reason from each of those two facts. It is harder than the stone, so movement now cracks the stone instead of the joint. And it is waterproof, so water that gets in behind it cannot leave through the joint and leaves through the face of the stone instead, carrying salts and freezing there in winter until the face flakes away. The joints survive perfectly and the building dissolves around them.
Worth knowing for completeness: the Romans had a mortar that set hard and even set under water, made by mixing lime with a volcanic ash from around Pozzuoli near Naples — which is why anything with that property is still called pozzolanic. Their harbour concretes are still in the sea. So "old" does not mean "weak" and "modern" does not mean "better". It means matched or not matched to the stone beside it.
A wall carries what is above it. Nothing else. So the stone at the top of a wall is carrying the coping — the capping course laid along the top to throw the rain off — and not much else, while the stone at the bottom is carrying the entire wall, plus the roof, plus the snow on the roof, plus the tower, plus the bell. The load is not the same everywhere. It accumulates all the way down, like the pressure at the bottom of a lake.
That alone would be reason enough to make the bottom stouter. But there is a better reason, and by now you can supply it yourself: the thrust line has to stay inside. A wall is never loaded perfectly evenly and never straight down — there is a roof pushing sideways, a wind, an arch springing from one face, a floor bearing on one side. Every one of those nudges the line off centre. The taller the wall, the more room the line needs by the time it reaches the ground, and the only way to give it more room is more thickness.
So walls taper, towers taper, lighthouses taper, chimneys taper, and a dry stone wall leans in as it rises. Not fashion. The line needs a wider track at the bottom, and the load is heaviest there.
The demonstration shows the second half of this module, which is one of the loveliest things in masonry. Cut a doorway into a wall and the load does not fall into the hole. It flows around it, arching over the opening through the mass of wall above and coming down through the stone on either side — the jambs, the vertical sides of the opening. A wall wants to arch over its own holes. Nobody has to tell it to.
Which means a lintel over a doorway is carrying far less than it appears to. It is not holding up the wall. It is holding up only the roughly triangular wedge of stone directly above it that the arching action leaves behind. And if that wedge is still too much for a stone beam — remember Module 4, and remember the underside — you can build a small arch into the wall just above the lintel, so the wedge is carried away as well. That is a relieving arch, and once you know the name you will see them everywhere, in brick above old windows and in stone above old doors.
The most famous version of the idea is older than the arch itself, and is not an arch at all. Above the Lion Gate at Mycenae, built around thirty-three centuries ago, sits an enormous stone lintel with a hollow triangle left open in the corbelled wall above it, and a carved slab of two lions set into the gap. The triangle does the same job by the same reasoning — it takes the wedge of load off the lintel and sheds it to the jambs — with corbelling instead of voussoirs, because nobody there had voussoirs yet. The sculpture is what everyone photographs. The triangle is the engineering.
You now have enough to walk up to a wall you have never met and hold an informed opinion about it. Masonry announces its troubles in a language of cracks and bulges, and every sentence in that language is about tension appearing where it should not.
And the quieter obligation, which is the same in every craft worth learning. Stone is the most patient material we build with and the least forgiving of hurry. Work you do to an old wall will outlast you by centuries, so the honest question is never "will this hold until I have gone?" but "what will the next person find, and will they thank me?" Match the mortar. Keep the water off. Do not lock in what ought to breathe. Leave it repairable.
You were not given a list of techniques. You were given one fact and then walked down its consequences, and the consequences turned out to be most of the history of building.
Stone is enormously strong when pushed and almost useless when pulled. Therefore stone must never be bent, because bending stretches one side of it. Therefore a flat stone lintel can only reach a short way across a gap, because its underside is being stretched, and the further it reaches the worse that gets. Therefore the arch: cut the stones as wedges and curve them, and the crossing becomes a chain of squeezes with nothing pulled anywhere. Therefore an arch must be built whole on a timber cradle, since half a chain of squeezes is no chain at all. Therefore we can ask where the squeeze runs, and the answer is the thrust line, which must stay inside the stone and preferably in the middle third of it. Therefore an arch shoves outwards at its feet as well as pressing down, since the curve that saved the stone is the same curve that turned some weight sideways. Therefore thick walls, buttresses, flying buttresses, and pinnacles that are nothing but useful weight bought to steepen a line. Therefore tie rods, when there is no more stone to spend — and pointed arches, which stand steeper and so shove less. Therefore the dome, which is an arch spun round, pressing outwards in every direction at once. Therefore hoop tension near a dome's base, and the vertical cracks it opens, and the iron ring that answers a ring of force with a ring of iron. Therefore mortar is a cushion that spreads the squeeze across imperfect faces, not a glue holding stones together — and therefore it should be softer than the stone, and therefore lime rather than cement in an old wall. Therefore a wall is thick at the bottom, because it carries everything above it and because the line needs a wider track by the time it lands. Therefore a wall arches over its own doorways, and a lintel carries only the wedge left over, and a relieving arch takes even that away.
Every one of those came from the first sentence. None of them had to be memorised.
So here is the test of whether it took. You are standing in front of a building nobody has explained to you. Why is that pier fatter than the others? What is that spike on the corner actually doing? Why is there a hollow triangle above that gate, and why does that crack step through the joints there but run straight through the stone over here? You were never taught the answers. You can derive them, because you hold the fact they all came from — and that is the difference between knowing a craft and having been shown one.
"As hangs the flexible line, so but inverted will stand the rigid arch."
— Robert Hooke, written in Latin as a scrambled
cipher in 1675 and unscrambled in print in 1705
The Guild of the Chisel · Course One of the Twelve Guild Courses · Iron Roots Supply
Back to the Guild Hall.
If this course sends you to one building, send yourself to the oldest arch you can walk to, and stand under it looking up at the joints.