A covenant for regenerative design.

The French materials chemist Joseph Davidovits coined the term geopolymer in the late 1970s and founded the Geopolymer Institute in northern France. Every modern alkali activated binder rests on the chemistry he established, and the field exists in its present form because of his work.

The larger claim came out of the same research. Studying the composition of stone at Giza, Davidovits proposed in 1979 that many of the blocks were not carved at all but cast in place, poured as a mineral slurry that set into artificial stone. He extended the argument to sites across South America, the Pacific and the Mediterranean, and spent four decades assembling the microstructural evidence behind it.

The response from mainstream archaeology has been to set the question aside rather than take it up. That is not difficult to understand. A finding that ancient builders synthesised their own stone would require rewriting the accepted account of how these monuments were made, over what timescale, and what the cultures that made them actually knew. The disciplines that would have to do the rewriting are the ones best placed to leave the question alone.

We build on his chemistry, and we take his archaeology seriously.

The case does not rest on one site, one method or one researcher. Independent analyses across two continents keep returning the same signature: chemistry that does not match the quarry, microstructure that is amorphous rather than crystalline, and surface behavior that belongs to a material which was once wet.

Three things at Cusco are visible without an instrument. Flow lamination runs continuously across a doorway lintel from one block into the next, which cannot happen in two stones cut separately from a quarry face. Trowel marks cross the exposed faces of the polygonal blocks, a finish that belongs to material worked while soft. And small protrusions sit at the bottom edge of block after block along the same course, positioned where fluid would collect and drain rather than where a rope would balance a load.

It would be tidier if this were a single tradition handed down from one civilisation to the next, and it is not. Giza precedes Roman concrete by more than two thousand years. Tiwanaku, Rapa Nui and Cusco all come long after Rome fell. There is no line to draw.

What there is instead is the same solution arriving separately, in places that had no contact with one another, whenever a culture understood its own ground well enough. Each one used what was locally to hand: Nile limestone, Neapolitan volcanic ash, Andean tuff and salt brine, Pacific basalt, Chinese clay and rice starch. Different ingredients, the same chemistry underneath. Each site answers a different question, and it is worth being precise about which.

Set them side by side and each one answers something the others cannot.

Age is answered at Giza. The casing has taken four and a half thousand years of Nile flooding and was stripped by an earthquake in 1303, and the cores are still standing. Rome adds nineteen centuries of an unreinforced dome held over a city. Tiwanaku adds fourteen.

Seismic survival is answered at Cusco, and it is answered better there than anywhere, because what matters in an earthquake is events survived rather than years elapsed. Five centuries is not long by the standards of this list. But in 1650 a quake destroyed the Church of Santo Domingo where it stood on the Qorikancha wall, and in 1950 more than half the city came down and destroyed the rebuilt church a second time, and both times the wall underneath did not move. Two direct hits, two collapses on top of it, no damage below. A thousand quiet years somewhere stable would tell us less.

The ocean is answered at Rapa Nui. In 1960 a tsunami generated off Chile crossed the Pacific and threw the moai at Ahu Tongariki inland. They were recovered and re erected, because being picked up by an ocean and put down somewhere else had not broken them.

Floods, earthquakes, tsunami, fire and four thousand years of weather, across five continents, in a material a modern engineer is told to be cautious about. No structure standing today that was built in the last hundred years is expected to do anything remotely like this.

Set the archaeology aside for a moment. The case for building in geopolymer today stands on measurements taken in the last twenty years.

Every culture in the section above did two things at once. They synthesised stone, and they aimed it at the sky.

Qorikancha's walls, gates and the streets around them were laid to meet the sunrise at the June solstice. At Machu Picchu a window is cut so that on the solstice morning a single ray crosses the room and lands on the stone at its center. In the Casma Valley, nine of twelve major sites face a solstice sunrise. These are not decorated buildings. They are instruments, and the decoration and the instrument are the same act.

That is the part of the inheritance we are most interested in. A material that lasts for millennia is only worth having if what you make from it deserves to last that long. Anyone can pour a durable box. The reason to learn to cast stone is that it lets you build something worth keeping, and then it keeps it.

Beauty is not the finish applied at the end when the budget allows. It is a decision made at the beginning, in the proportions, in where the openings go, in what the light does at four in the afternoon in November. Those decisions cost nothing extra. They only require that somebody cared enough to make them on purpose.

The oldest design tools are ratios rather than dimensions. Vitruvius built his whole theory on commensurability, the idea that every part of a building is a whole number fraction of the whole. The Egyptians measured slope with the seked, the horizontal run for a rise of one cubit, and we know this because they wrote it down in the Rhind and Moscow papyri.

The Great Pyramid's faces resolve to a seked of five and a half palms per cubit. That works out at 51.84 degrees, which happens to put the slant and the half base into golden ratio with each other. Whether the builders were reaching for that number or arrived at it through a system of whole palms is argued about and probably unanswerable.

The more interesting possibility is that it does not matter. A builder working in whole units, with a cord and a plumb bob, arrives at the proportion without naming it. It is in the method rather than in the theory.

What we take from this is a working rule. Decide the ratio system before deciding the shape, keep it consistent from the site plan down to the sill of a window, and let form come out of it rather than be applied to it. That is what makes a building feel inevitable rather than arranged.

Orientation is free. It is decided once, at the start, by turning the plan a few degrees, and then it works every day for as long as the building stands.

A room can be laid so the winter sun reaches the back wall and the summer sun does not. An opening can be cut to catch a particular sunrise once a year, or the moon at a standstill, or nothing at all except the way the light moves across a floor between breakfast and noon. The Andean builders did this at monumental scale with the solstices. It works just as well at the scale of one room and one person standing in it.

A ridge crown with a long horizon is unusually good ground for this, because you can actually see where the sun comes up across the year rather than guessing at it behind a hillside.

The practical shape of the work is closer to earthbuilding than to a concrete pour. Precursor powder is dry blended. Activator solution is batched separately, in a shaded station with gloves, eye protection and running water, because concentrated hydroxide is caustic until it is in the mix. The two meet in a mixer and the batch has a working window measured in tens of minutes rather than hours, so forms are ready and hands are counted before anything is wetted.

Placement is by wheelbarrow and bucket into standing forms, rodded or vibrated to close the voids against the face. The pour is levelled, covered to keep moisture in, and left. Ambient cure is slower than the heated cure used in laboratories, so a wall is not stripped the next morning: initial set within the day, forms moved after a few days, and strength climbing for weeks afterwards. That pace suits a resident crew far better than a contractor's schedule, because the wall does its work while the people who poured it are doing something else.

Nothing here needs a factory. It needs a mixer, forms, shade, water, protective gear and a group of people willing to come back tomorrow.

Removing steel removes tensile strength, and the way it comes back is fiber dispersed through the mix. Basalt fiber is drawn from melted basalt rock, needs no additives, does not corrode, and is a material we would be casting beside anyway.

A 2021 study in Sustainability ran fly ash based geopolymer mortar filled with waste basalt powder, at four fiber dosages, through twenty eight day strength tests and then through 390, 750, 1,110 and 1,470 °F (200 to 800 °C). The pattern is consistent. At the highest dosage, compressive strength rose about 12 percent and flexural strength about 34 percent before any heating. Impact resistance more than doubled, up 118 percent. After exposure to 1,470 °F (800 °C), the fiber specimens held 16 percent more compressive strength and 45 percent more flexural strength than the plain mix, lost less weight, and showed measurably less cracking.

The study is honest about the cost, and so are we: more fiber raised capillary water absorption by roughly 18 percent, because fiber opens up pore connectivity. In a wall that means detailing matters, a good roof and a dry base rather than a different material.

Fiber is how a wall with nothing to rust inside it still bends a little before it breaks.

Cob, adobe, rammed earth, hempcrete and straw bale all deserve their revival. They are cheap, they are gentle to make, they can be built by hand, and people are living well in them right now. We have no argument with any of them, and the comparison below is not an attempt to win one.

What separates them is the binder. Earthen walls are held together by clay, which is a physical bond that water can undo. Hempcrete and straw bale are held together by lime around a plant fiber, and the plant is the part that eventually goes. Geopolymer is held together by a mineral network that formed chemically and does not dissolve, does not burn, and has nothing organic in it to decay.

The trade is honest: earthen materials cost less to make and more to keep. They want a good hat and good boots, a generous roof overhang, a raised plinth and a render that gets renewed on a cycle measured in years. That maintenance is a practice, and for many people it is a welcome one. Our reason for choosing otherwise is that the structures we want to build should not require it.

Two different questions get confused here, so we will separate them.

Per cubic yard, geopolymer is not the cheap option, and the honest answer is that nobody can give you a single number. Published comparisons run from about seven percent cheaper than Portland cement to roughly double it. Almost all of that spread is the alkaline activator, which is sold in different forms, grades and purities at wildly different prices, so what you pay depends far more on how and where you buy than on the chemistry. Waste stream precursors and higher strength per unit close part of the gap. Bulk sourcing closes more of it. Expect a premium and expect it to be modest rather than crippling.

The saving is not in the material. It is in everything the material makes unnecessary. A geopolymer structure carries no steel, so it has no corrosion clock and no replacement cycle for reinforcement. Its own mass flattens the daily temperature swing, which removes most of the reason a building here needs a furnace or a heat pump, and with them the ducting, the equipment replacement cycle and the bill. It does not burn. And it is built by the people who will live in it, on land the foundation already holds, without a mortgage, so the interest that costs more than the house itself over thirty years is simply absent.

That is the comparison that matters, and it is not a comparison between two prices per cubic yard. We will put real figures here when a prototype has produced them.