Megaprojects of Antiquity: The Logistics of Moving Millions of Tons Without Steel

The oldest inscribed papyri ever found are not hymns or king lists. They are a logbook. In 2013, at Wadi al-Jarf on Egypt's Red Sea coast, a team led by Pierre Tallet pulled out the records of a mid-level inspector called Merer, who spent months of Khufu's reign moving limestone by boat from the quarries at Tura to the building site at Giza. If you want to know how the Great Pyramid got built, the most important primary source we have is, more or less, a shift report.

That fact should embarrass a whole genre. The pitch for this article is that we would still struggle to repeat what the ancients did without heavy machinery, and the usual way that pitch gets cashed out is mystery: lost techniques, impossible precision, secrets in the stone. I think both halves are wrong. A modern contractor with cranes and trucks could do it, and what made it hard in the first place was not technology anyone has lost. It was food, water, boats, scheduling and the patience to survey a line to the centimetre. The monuments are what logistics looks like when it is allowed to run for decades.

§A middle manager's diary

Merer's entries are short and repetitive, which is exactly what makes them valuable. One of them covers a three-day run: load at Tura, overnight at a stop called She-Khufu, arrive at Akhet-Khufu, the name for the pyramid site, then back to Tura for the next load. He reports to Ankhhaf, Khufu's half-brother, who carried the title of head of all the king's works. The diary names artificial basins at Giza, a Lake of Khufu among them, flooded in the high-water season so that boats could come right up to the construction zone. Another set of papyri from the same cache records six months spent shipping workers, food and copper across the Red Sea.^[1]1. On the find and the contents of the diary, National Geographic's account of the Red Sea scrolls; the full story is in Pierre Tallet and Mark Lehner, The Red Sea Scrolls (Thames & Hudson, 2021).

Read that as an operations document and it is startlingly familiar. There is a recurring route, a fixed cadence, a named waypoint, a reporting line that goes up to the person accountable for the whole programme, and a separate supply operation bringing in the copper the stonecutters wore down. Wadi al-Jarf itself was a harbour built for the Sinai copper runs: a stone jetty, anchors by the hundred, storage galleries cut into the hill. Nothing here is mysterious. It is a supply chain with a monument at the end of it.

§The river did the lifting

The single biggest piece of technology at Giza was the Nile. The limestone for the casing came from across the river, the granite from Aswan far to the south, and both moved by water for the long legs. In 2022 Hader Sheisha and colleagues used pollen preserved in sediment cores to reconstruct eight thousand years of water levels on the Giza floodplain. They found that during the reigns of Khufu, Khafre and Menkaure, a now-vanished arm of the river they call the Khufu branch still stood at roughly 40 per cent of its Holocene maximum, high enough to bring boats to the harbour complex below the plateau.^[2]2. Hader Sheisha et al., Nile waterscapes facilitated the construction of the Giza pyramids during the 3rd millennium BCE, PNAS 119 (2022), e2202530119. The pyramids sit where they sit partly because that is where the barges could reach.

For the last few hundred metres, where boats could not go, you drag. Here the best-known modern result is a nice piece of physics. In 2014 Daniel Bonn's group in Amsterdam put a model sledge on a tray of sand and measured the pulling force as they added water. Dry sand heaps up in front of the runners. A little water forms capillary bridges between the grains, roughly doubles the sand's stiffness, and stops the heap forming; at the right moisture the force needed drops by about half. Add too much and the bridges merge and the effect goes away.^[3]3. A. Fall, B. Weber, M. Pakpour, N. Lenoir, N. Shahidzadeh, J. Fiscina, C. Wagner and D. Bonn, “Sliding Friction on Wet and Dry Sand”, Physical Review Letters 112, 175502 (2014); summaries from APS Physics and the University of Amsterdam.

Sliding friction is close enough to linear that the crew size scales with it:

Halve $\mu$ and you halve the number of people on the rope, which means halving the bread, beer and housing for that part of the job. That is the real payoff, and it is a logistics payoff, not a mechanical one.

Two cautions, because this result gets oversold. The painting that inspired it, in the tomb of Djehutihotep, shows a man pouring water in front of a sledge carrying a colossal statue, and it dates to the Middle Kingdom, centuries after Khufu. It is good evidence that Egyptians knew the trick. It is not a photograph of the Great Pyramid under construction. And the experiment was a lab model, so the “half the workers” figure is an upper bound on a favourable surface, not a measured fact about Giza.

§Who was on the rope

The most durable myth about the pyramids is that slaves built them. It goes back to Herodotus, writing roughly two thousand years after the fact, and Hollywood did the rest. The archaeology does not support it.

South of the Sphinx, Mark Lehner's team at AERA has excavated about ten acres of a Fourth Dynasty settlement now called Heit el-Ghurab: barracks, granaries, bakeries, scribal offices, cattle pens and harbour basins. Lehner has come to see it as a major Nile port as well as a workers' town. The barracks could sleep 1,600 or more, and the rubbish heaps are full of cattle, sheep, goat and fish bones. A cemetery for the workers, found in 1990, holds modest tombs with goods for the afterlife, right beside the monuments they built. Inside the pyramids, work gangs left their names as graffiti: the Followers of the Powerful White Crown of Khufu, the Drunkards of Menkaure.^[4]4. On Heit el-Ghurab, see AERA's page on Mark Lehner; on the barracks, the workers' cemetery and the gang names, Eric Betz, Who Built the Egyptian Pyramids? Not Slaves, Discover (2021).

Nobody names their crew the Drunkards of Menkaure under the lash. That is the name of a team with an identity, a rivalry and probably a favourite bar. But I want to be careful not to swing to the opposite myth, of happy volunteers. The likeliest model is labour as a form of tax, with people rotating through the works for a term, fed and housed by the state. That is compulsion. It is not chattel slavery, and the difference shows up in exactly the places the evidence lives: the rations, the graves, the names.

From an engineer's seat, the interesting thing about the workers is not their legal status but the provisioning. Every person on site needs bread, which means grain, which means granaries and ovens and water and fuel, every day, for years. Whatever the size of the workforce, the bakery was as much a part of the build as the ramps.

§A quarter of a metre per kilometre

Rome is the counterexample to the idea that ancient megaprojects ran on muscle alone. Their aqueducts ran on surveying.

The aqueduct that fed Nîmes, built around 50 CE, runs about 50 kilometres from a spring near Uzès, and it delivered 30,000 to 40,000 cubic metres of water a day by gravity. Its average fall, according to the site itself, is 25 centimetres per kilometre. The Pont du Gard, the bridge that carries it over the Gardon, stands 49 metres high, and across the bridge the channel drops by about 2.5 centimetres.^[5]5. Figures from the site's own history page. The 2.5 cm across the bridge is from M. J. T. Lewis, Surveying Instruments of Greece and Rome (Cambridge, 2001), p. 187, as cited in Wikipedia's Pont du Gard article, which also gives A. Trevor Hodge's figures for how unevenly the gradient varies along the route. Put those as ratios:

At the average rate the whole route falls only about twelve and a half metres end to end. The gradient is also not uniform; it is steeper in some stretches and almost flat in others, which is what you would expect from engineers fitting a line to real terrain rather than drawing one on a plan.

What makes this more impressive is the instrument. Vitruvius describes the chorobates, a levelling table about twenty feet long with plumb lines down its legs, and, for windy days, a groove cut in the top to be filled with water and read like a spirit level. He also recommends that an aqueduct fall not less than half a foot in a hundred.^[6]6. Vitruvius, De Architectura 8.5–8.6, in the LacusCurtius translation. The text does not repeat the unit after “one hundred”, and the mismatch with real aqueducts has puzzled scholars for a long time. That is one in two hundred. The Nîmes builders went twenty times flatter than the textbook. Whether that was confidence, necessity forced on them by the terrain, or a corrupt line in the manuscript, the outcome is the same: a chain of crews with wooden instruments held a slope of a quarter of a metre per kilometre across hills and valleys for fifty kilometres.

This is the place where “we'd struggle to replicate it” is most wrong, and most interestingly so. A modern survey crew with a laser level would do this in an afternoon. The Roman achievement was not that the measurement was beyond us. It was that they turned a fragile measurement into a repeatable procedure, and ran it, segment by segment, through a workforce the site puts at nearly a thousand men over about five years. Precision as a process, not a talent.

§At Angkor the canal was the crane

Angkor Wat, built in the twelfth century, holds somewhere between five and ten million sandstone blocks by one widely reported estimate, some of them up to about a tonne and a half. For a long time the assumed route from the quarries was a roundabout river journey of about 90 kilometres. In 2012 Etsuo Uchida and Ichita Shimoda of Waseda University used satellite imagery and fieldwork to trace something better: more than fifty quarries along the foot of the Kulen hills, linked to the temples by a canal route of roughly 35 kilometres. Some of the canal sections still hold water, and stray sandstone blocks along the way look like they fell off rafts in transit.^[7]7. Etsuo Uchida and Ichita Shimoda, “Quarries and transportation routes of Angkor monument sandstone blocks”, Journal of Archaeological Science (2012); reported by Live Science via NBC News and Southeast Asian Archaeology, which give slightly different route lengths.

The canals were not a one-off. The Greater Angkor Project's map of the region covers nearly 3,000 square kilometres and shows a water management network spread over more than 1,000 of them: channels, embankments and enormous reservoirs that stored the monsoon and spread it across a low-density city.^[8]8. Damian Evans et al., A comprehensive archaeological map of the world's largest preindustrial settlement complex at Angkor, Cambodia, PNAS 104 (2007), 14277–14282. The temples are the part tourists photograph. The hydraulic system is the part that let a city of that size exist at all, and that delivered the stone.

There is a sting here too. The same paper argues that the scale of landscape modification was large enough to cause real damage, deforestation, erosion and degraded topsoil, and supports the old hypothesis that overexploitation of the landscape played a part in Angkor's decline. The network that built the city also became the thing the city could not afford to maintain. Every infrastructure engineer has met a smaller version of that problem.

§We could. We would just go over budget

So could we replicate these projects? Physically, easily. A modern contractor would cut the stone with diamond wire, move it by truck, set it with a crane and survey it with lasers. Nobody needs a lost secret.

The harder question is whether we could replicate the coordination, and there I think the hook has a point it does not know it is making. Bent Flyvbjerg, who has spent a career collecting data on modern megaprojects, summarises the pattern as an iron law: over budget, over time, over and over again.^[9]9. Bent Flyvbjerg, What You Should Know About Megaprojects, and Why: An Overview (2014), published in the Project Management Journal. Our failures are almost never about the physics. They are about estimating, incentives, interfaces between contractors, and the long dull work of keeping thousands of people supplied and pointed in the same direction. Those are exactly the problems Merer, the Nîmes surveyors and the Angkor canal builders were solving, without spreadsheets, over decades.

They also had advantages no modern project gets, and it would be dishonest to skip them. A god-king does not face a planning inquiry. A state that can draft labour as tax does not negotiate wages. Time horizons of a reign or longer make a twenty-year build normal rather than scandalous. Some of what looks like superior coordination was simply the absence of anyone allowed to say no.

Which is why I keep coming back to Merer. He is the most modern person at Giza: a man with a route, a crew, a quota and a boss, writing down every trip because someone above him would want to know. The pyramid has had four and a half thousand years of people staring at it and asking how. His logbook answers in the least romantic way possible. Boats, mostly. And someone keeping count.