What Makes a City a City

Çatalhöyük, in central Anatolia, held several thousand people around 7000 BCE. The houses were packed so tightly there were no streets; people walked across the roofs and entered through holes in them. It is one of the largest settlements of its era, it is spectacular, and it is not a city.

There are no public buildings. No storehouse, no plaza, no temple distinct from a dwelling. Each household made its own tools, ground its own grain, buried its own dead under its own floor. Nobody appears to have specialised in anything, and nobody appears to have told anybody else what to do. Çatalhöyük is a very large village, which is a different kind of thing.

What makes a city is not size. It is the presence of people who do not produce their own food, and the machinery that feeds them.

Once a settlement contains full-time potters, smiths, scribes, soldiers and priests, grain has to flow from the countryside to them, and that flow must be assessed, collected, stored, guarded and distributed. This is why cities and administration appear in the archaeological record together and never separately: the specialists are only possible because the accounting exists, and the accounting is only necessary because the specialists do.

The place it happens first is southern Mesopotamia, and the reasons are specific enough to be worth stating.

The southern floodplain is unpromising at first glance: barely any rainfall, punishing summers, and rivers that flood at the wrong time of year. What it has is silt, flat ground and water that can be led onto fields. Irrigated barley there produced yields that rain-fed farming could not touch, and irrigation demands coordination (canals must be dug and dredged, water allocated, disputes settled) which concentrates both food and authority in the same hands.

But the alluvium has nothing else. No stone, no timber, no copper, no tin. Every building material beyond mud, every tool beyond clay and reed, had to come from somewhere else, which meant that long-distance exchange was not a refinement of urban life in Mesopotamia but a precondition of it. Cities there were nodes in a trading system before they were anything else, and the records they kept were largely about that.

James C. Scott added a sharper question in Against the Grain: why is every early state in the world a grain state?

Wheat, barley, rice, millet, maize: never cassava, never potatoes, never yams, although those crops feed people perfectly well and in some cases better. Scott's answer is that a state needs a crop it can see. Grain ripens on a schedule, stands above ground, ripens all at once, and can be counted, divided, stored and transported. A tax assessor can stand at the edge of a field and estimate the yield. Tubers sit in the ground until you dig them, which means they can be harvested a little at a time, as needed, by people who would rather not be assessed.

On this reading the grain state is not a fact about nutrition but a fact about legibility, and the earliest states are built on the crop that is easiest to take.

That framing has an uncomfortable corollary about how cities filled up. The romantic version has people drawn in by opportunity. The demographic arithmetic makes that impossible on its own: urban death rates exceeded birth rates almost everywhere until the nineteenth century, so a city that is not receiving a constant inflow shrinks. And a great deal of that inflow was not voluntary. Mesopotamian records document war captives settled on the land, debt bondage, and whole populations resettled by decree. Walls hold people in as well as out.

Cities were genuinely attractive to many people. They were also, in their first millennia, substantially populated by people who had not chosen to be there.

One case refuses to fit any of this, and it is worth ending on because it marks the edge of what we actually know. The Indus civilisation, contemporary with early Mesopotamia and larger in area, built Mohenjo-daro and Harappa on grids, with standardised fired bricks, covered drains, public baths and weights that are consistent across hundreds of miles.

And it has produced no palaces. No royal tombs. No unmistakable portrait of a ruler, no monument to a conquest, no scene of a king receiving tribute. Nothing of the iconography that saturates every Mesopotamian and Egyptian site of the period. Something standardised brick dimensions across a region larger than either. Whether that something was a state, and what kind, remains open, and it will remain open as long as the script does. About four thousand short inscriptions survive, nobody has deciphered them, and there is no bilingual text.

The most sophisticated urban civilisation of its millennium is, politically, mute.