Seasonal labour budget

Why were the greatest ancient monuments built in the months when the fields needed nobody?

The usual explanation for a pyramid is power: a king commanded, and the stone moved. But power does not answer the interesting question, which is not who but when. Egypt's great building campaigns cluster into the months of the inundation, when the Nile spread across the fields and there was nothing agricultural to do. So do medieval labour services, and the Andean mit'a, and a great many other large works in agrarian states.

If the state could simply command, why did it wait for a particular season? A king who can compel labour in August can compel it in April. Unless something about April is not his to spend.

Try treating labour the way a physicist treats a conserved quantity, and see how far it carries.

A village of a thousand adults has, in a year, a fixed stock of person-days. Not approximately fixed — genuinely fixed, because a person cannot work two days in one. Every day spent quarrying is a day not spent farming, and no decree changes the total. So the question "how do we build a pyramid?" is not a question about will. It is an accounting question: which entries in a closed ledger get moved?

Now ask what the agricultural entries look like across the year. Here is the crucial observation, and it is easy to miss: farming demand is not steady. It is spiky. Ploughing and sowing need every hand within a narrow window set by the water and the temperature; harvest needs even more hands in an even narrower window, because grain left standing is grain lost. Between those spikes are long stretches when the crop grows by itself and human labour adds nothing at all.

So the annual labour supply is flat and the agricultural demand is peaked. Subtract one from the other and you have not a shortage but a shape — a trough of person-days that the fields cannot use and that the year cannot store. Labour does not keep. A day of a farmer's time in August, unused, is simply gone.

Now the ledger explains the calendar. Building in the trough costs the food supply nothing, because those days were producing no food anyway. Building at harvest costs it everything, since each person-day taken is a person-day of grain not gathered. The monument is not paid for out of surplus wealth. It is paid for out of the unevenness of the farming year — a quantity that would otherwise be discarded.

There is a second conserved account running alongside, and it constrains the first. Workers eat whether or not they farm. Feeding twenty thousand people at a building site requires grain that already exists, in a granary, from a previous harvest. So the real precondition is storage: a state can convert idle time into stone only to the extent it can hold calories across months. That is why monumental building and centralised granaries appear together so persistently — neither is much use without the other.

And in Egypt the flood did something further that is worth naming, because it is a genuine bonus rather than an inference. The rising water reached inland, so heavy stone could be floated most of the way to the site. The papyri found at Wadi al-Jarf — the logbook of an official named Merer, from the twenty-sixth year of Khufu's reign — record boat crews ferrying limestone from the Tura quarries to Giza. The same event that emptied the labour ledger also cut the transport cost.

One correction to the tidy story, since the tidy story is now known to be too tidy. The picture of pyramids built entirely by flood-season peasants has been substantially revised by the excavation of the workers' settlement at Giza. That site shows a permanent, fed, skilled workforce — bakeries, breweries, an enormous supply of cattle and sheep — with conscripted gangs rotating through alongside it. Modern estimates put the total nearer twenty thousand than the hundred thousand Herodotus reported. So the seasonal logic set the rhythm of the mass labour, but it did not do all the work.

The natural misreading is that these societies had "spare" labour in some absolute sense, as though there were idle hands lying about. There were not. What existed was a timing mismatch: the same hands were desperately needed in two short windows and superfluous in between. Take the same total labour and flatten its agricultural demand — as continuous irrigated cropping or double-harvest regimes tend to do — and the trough shrinks, and with it the seasonal building window.

The second refinement concerns coercion. The conservation argument does not say the labour was voluntary or that corvée was gentle; it was a tax, paid in days rather than grain, and evasion was punished. What it says is narrower and more interesting: coercion sets who pays, but the ledger sets how much there is to take. A king who conscripts at harvest is not being more powerful than one who conscripts at flood. He is eating his own granary a year early, and the following season will present him with the bill.

• How storage technology set the ceiling — what a granary's capacity and spoilage rate imply about the largest project a state could attempt.
• Why the Inca mit'a taxed time rather than produce, and what that reveals about a state with no money and no draught animals.
• What changed once wage labour and food markets let building be scheduled by price rather than by season.

Every term the collection defines is gathered in the glossary.