Khipu: The Empire That Kept Its Books in Knots

Read the episode (Narration script)

Welcome to Origins. In this program we explore the technological and scientific roots of the modern world. Let's begin.

In the late 1540s, a young Spaniard named Pedro de Cieza de León was traveling through Peru, filling notebooks. He kept hearing the same improbable thing. The Inka, people said, had run an empire of millions without writing, and every province had kept accounts so exact that, years later, not so much as a pair of sandals would go missing.

Cieza didn't believe it. He wrote that he held most of the story to be a fable.

So in the valley of Xauxa, in the central highlands, he asked a local lord named Guacarapora to prove it. Guacarapora sent his servants to fetch the records. They came back carrying bundles of string.

The lord went through them with him. Everything his people had handed over to the Spaniards since Francisco Pizarro first came into the valley was there: the gold, the silver, the cloth, the maize, the flocks, all of it noted without fault or omission. Cieza wrote that he was quite astonished.

What he was looking at was a khipu. The word means "knot" in Quechua, and the Spanish usually wrote it quipu.

Picture a main cord, a few inches to several feet long. Hanging from it, like the fringe on a shawl, are pendant cords of cotton or llama and alpaca wool, often dyed. Some khipus have a handful of pendants. The largest have around fifteen hundred. Knots sit along the pendants in clusters, and some pendants carry smaller cords of their own, branching off like twigs.

The Inka state was the largest in the Americas when the Spanish arrived, stretched along thousands of miles of the Andes. It moved labor, food, cloth and soldiers across that distance and stocked storehouses along its roads. That takes bookkeeping. The bookkeepers were called khipukamayuq, the keepers of the cords. According to Cieza's sources in Cuzco, every province knotted its births and deaths onto khipus each year and carried them to the capital, where the whole empire was tallied.

The Inka didn't invent the technique. Khipus survive from the Wari, an Andean state that flourished centuries earlier, in a region where spinning and weaving were among the most prestigious crafts a person could practice.

One of the best witnesses to how khipus were used was a boy. Garcilaso de la Vega was born in Cuzco in 1539, the son of a Spanish captain and an Inka noblewoman. In a history published in 1609, he remembered how local lords came into the city at the feast of Saint John in June, and again at Christmas, to pay their tribute. They brought Spanish written copies of what they owed, along with their own knots, and they asked his mother to have the boy check one against the other. They didn't trust the Spaniards to tell them what the paper said. So a child who could read both systems sat between them, reconciling a ledger in ink with a ledger in string.

Then the Church moved against the cords. In 1583, the Third Council of Lima ordered the destruction of khipus that held records of the old religion, calling them pernicious. The same council encouraged priests to have converts make khipus of their sins for confession. Khipus kept working in village accounts and even colonial lawsuits, but the people who could read them grew fewer each generation.

By the twentieth century, the surviving khipus sat in museum drawers in Lima, Berlin and New York, and no one could say with authority what they meant. Some writers suspected they were memory aids, like rosary beads, meaningless without a speaker.

The first real foothold came from a mathematics teacher in Brooklyn. Leland Locke had studied the history of arithmetic at Columbia, and he got access to a collection of Peruvian khipus at the American Museum of Natural History. In 1912 he published his reading of one, from Huando, a coastal village north of Lima.

Locke saw that the knots sat in rows at set distances from the main cord. The row farthest from the main cord held the ones. Above it, the tens, then the hundreds, then the thousands. A single knot in the hundreds row meant one hundred. A position with no knots meant zero. It was a base-ten, place-value number system, made of fiber.

The proof was on the specimen itself. Its pendants were gathered in groups, and each group had one extra cord looped across their tops. When Locke added up the knots on the pendants, the total matched the knots on that top cord. The khipu-maker had left a checksum.

That settled the numbers and raised a harder question. A column of figures needs labels. Twenty-seven of what? Belonging to whom?

In 2005, the anthropologist Gary Urton and his collaborator Carrie Brezine looked at twenty-one khipus from Puruchuco, an Inka administrative center near Lima, using a database of hundreds of cords. Seven of them fit together. Values on one khipu were summed onto another, and those totals summed again, the way local accounts roll up into regional ones. It was the shape of a bureaucracy, tied in string.

The labels, it seems, live in color, in fiber, in the direction a cord was spun, and in whether a pendant was tied to the front or the back of the main cord.

In the mid-2010s, a Harvard undergraduate named Manny Medrano spent his freshman spring break comparing six khipus from Peru's Santa Valley with a Spanish census, taken in 1670, of about a hundred and thirty-two tribute payers. The way each cord was attached seemed to record which half of the community a person belonged to, and the colors seemed to track their names. It is still the only known case where a set of khipus can be checked against a matching written census.

High in the Andes, the village of San Juan de Collata kept two khipus in its own archive. In 2015 it invited the anthropologist Sabine Hyland to study them. Villagers said they were letters, sent between communities during an eighteenth-century revolt against Spanish rule. The cords use fourteen colors and six animal fibers, including vicuña, alpaca and deer. Village authorities handled them without gloves, because some fibers are easier to tell apart by touch than by sight. Hyland counted ninety-five distinct signs and argued that some stand for syllables. Other specialists want more examples before calling it phonetic writing.

About nine hundred khipus are known to survive, and most remain unread. What has been recovered so far is enough to widen the idea of writing. A civilization stored counts, categories, and perhaps names in twist, color, texture and position, and ran a state on the result. It was a record made to be handled as much as looked at.

Cieza came expecting a fable. He found a man holding the receipts.

That's our show for today. Thanks for listening to Origins.