The Quipu: An Empire Written in Knots

Two honest qualifications belong here. First, the numbers this system tracked are staggering but imprecise to us: the Inca population is often given as ten to twelve million, but that is really the middle of a wide and genuinely disputed range, with serious estimates running from around four million to more than thirty million depending on method, so the figure should be held loosely rather than quoted as fact. Second, color clearly mattered and clearly carried meaning, with different colors of cord lining up with different categories of thing counted, maize against potatoes against workers, but no universal color key has ever been recovered. The meanings seem to have shifted by region and era, and perhaps by the lineage of the knot-keeper, which is one reason no single Rosetta Stone for the quipu has ever been found.

Which brings us to the real frontier, and the question that makes the quipu more than an abacus. Did the cords also record language, not just quantity? There are serious reasons to think they might. The anthropologist Gary Urton has argued that each knot involves a string of yes-or-no choices, of fiber and color and spin and knot type and direction, seven such choices in all, which would give a hundred and twenty-eight possible combinations per knot, enough, in principle, to encode syllables the way a handful of binary bits encodes a letter. Sabine Hyland, working with two 'narrative' quipus that villagers in San Juan de Collata had preserved for generations as the record of a rebellion, found ninety-five distinct cord-signs, consistent with a syllabic script for the local dialect. And Manny Medrano, as a Harvard undergraduate working with Urton, matched six khipus from the Santa Valley to a 1670 Spanish census and showed that the direction a cord is attached to the main cord, whether it comes off the front or the back, tracks which of two social groups a person belonged to, while the cords' colors seem to line up with their first names. These are real, specific, peer-reviewed advances, and they are genuinely exciting. But they must be held for exactly what they are. Each decodes a feature or two on a handful of particular cords tied to a known document; none of it is a general ability to read a quipu cold. The field's own honest verdict, which this article keeps, is blunt: no one can currently read a quipu the way one reads a page of text. The narrative question is not closed. It is open, and moving, and unsolved.

It is worth pausing on why so little is left to read. When the Spanish understood that the quipu was the backbone of Inca governance, and suspected it of preserving a religion they meant to erase, they set out to destroy it. The Third Council of Lima ordered quipus burned in 1583; the viceroy Francisco de Toledo replaced the knot-keepers with Spanish notaries. The destruction was thorough but not total: roughly nine hundred quipus survive today, scattered across museums, a small fraction of what once existed. And there is a cruel possibility folded into that number. If some quipus really did record history and narrative rather than tallies and totals, those are exactly the ones the Spanish would have most wanted gone, which means the surviving sample may be skewed toward the account books and away from the stories, and the richest evidence for the very question we most want to answer may be the evidence most deliberately lost.

Two opposite errors are worth refusing outright. The first, older and more insulting, is that the quipu was a 'primitive' counting toy, a knotted tally and nothing more. The evidence flatly contradicts it: a system that ran the census and tribute of more than eighty provinces, with hierarchical cords, categorical color, and multi-feature construction, is a genuine information technology, whatever else it may or may not turn out to be. The second error runs the other way: that the quipu has now been 'fully deciphered' as writing, or that the narrative-reading question is settled. It has not been, and it is not. The numbers are read; the language, if it is there at all, is not, and the honest, exciting truth is that we do not yet know how much a quipu can say. One last tidy story deserves refusing too: that the quipu must have been borrowed from some Old World knotted-string tradition, Chinese or Polynesian. It need not have been. Recording information in knotted cord is the kind of good idea different peoples reach independently, and there is no evidence of any contact. The quipu is an Andean invention, and its unsolved half is an Andean secret still.